finalizing 9.4.1 release

use adc array,y or sbc array,y instead of tempvar

.github/workflows/all-ci.yml

@@ -10,7 +10,7 @@ jobs:
 
     steps:
       - name: Checkout code
-        uses: actions/checkout@v2
+        uses: actions/checkout@v3
 
       - name: build and install recent 64tass
         run: |
@@ -21,7 +21,7 @@ jobs:
           sudo make install
 
       - name: Set up JDK 11
-        uses: actions/setup-java@v2
+        uses: actions/setup-java@v3
         with:
           java-version: 11
           distribution: adopt

.gitignore

@@ -15,6 +15,7 @@ out/
 parser/**/*.interp
 parser/**/*.tokens
 parser/**/*.java
+compiler/src/prog8/buildversion/*
 *.py[cod]
 *.egg
 *.egg-info
@@ -29,6 +30,8 @@ parsetab.py
 compiler/lib/
 
 .gradle
+**/BuildVersion.kt
 /prog8compiler.jar
 sd*.img
 *.d64
+

.idea/kotlinc.xml

@@ -4,6 +4,6 @@
     <option name="jvmTarget" value="11" />
   </component>
   <component name="KotlinJpsPluginSettings">
-    <option name="version" value="1.8.21-release-380" />
+    <option name="version" value="1.9.0-release-358" />
   </component>
 </project>

.idea/libraries/io_kotest_property_jvm.xml

@@ -1,24 +0,0 @@
-<component name="libraryTable">
-  <library name="io.kotest.property.jvm" type="repository">
-    <properties maven-id="io.kotest:kotest-property-jvm:5.6.2" />
-    <CLASSES>
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-property-jvm/5.6.2/kotest-property-jvm-5.6.2.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/com/github/curious-odd-man/rgxgen/1.4/rgxgen-1.4.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/1.8.10/kotlin-stdlib-jdk8-1.8.10.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/1.8.10/kotlin-stdlib-1.8.10.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/1.8.10/kotlin-stdlib-jdk7-1.8.10.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-common-jvm/5.6.2/kotest-common-jvm-5.6.2.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-shared-jvm/5.6.2/kotest-assertions-shared-jvm-5.6.2.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-api-jvm/5.6.2/kotest-assertions-api-jvm-5.6.2.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-jdk8/1.7.0/kotlinx-coroutines-jdk8-1.7.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/opentest4j/opentest4j/1.2.0/opentest4j-1.2.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-common/1.8.10/kotlin-stdlib-common-1.8.10.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/github/java-diff-utils/java-diff-utils/4.12/java-diff-utils-4.12.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-reflect/1.8.10/kotlin-reflect-1.8.10.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-core-jvm/1.7.0/kotlinx-coroutines-core-jvm-1.7.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/annotations/23.0.0/annotations-23.0.0.jar!/" />
-    </CLASSES>
-    <JAVADOC />
-    <SOURCES />
-  </library>
-</component>

.idea/libraries/michael_bull_kotlin_result_jvm.xml

@@ -1,8 +1,8 @@
 <component name="libraryTable">
   <library name="michael.bull.kotlin.result.jvm" type="repository">
-    <properties maven-id="com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.17" />
+    <properties maven-id="com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.18" />
     <CLASSES>
-      <root url="jar://$MAVEN_REPOSITORY$/com/michael-bull/kotlin-result/kotlin-result-jvm/1.1.17/kotlin-result-jvm-1.1.17.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/com/michael-bull/kotlin-result/kotlin-result-jvm/1.1.18/kotlin-result-jvm-1.1.18.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/1.8.10/kotlin-stdlib-jdk8-1.8.10.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/1.8.10/kotlin-stdlib-1.8.10.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/annotations/13.0/annotations-13.0.jar!/" />

README.md

@@ -14,6 +14,21 @@ Documentation
 Full documentation (syntax reference, how to use the language and the compiler, etc.) can be found at:
 https://prog8.readthedocs.io/
 
+How to get it/build it
+----------------------
+
+- Download the latest [official release](https://github.com/irmen/prog8/releases) from github.
+- Or, if you want/need a bleeding edge development version, you can:
+  - download a build artifact zipfile from a recent [github action build](https://github.com/irmen/prog8/actions).
+  - you can also compile it yourself from source. [Instructions here](https://prog8.readthedocs.io/en/latest/compiling.html).
+
+Community
+---------
+Most of the development on Prog8 and the use of it is currently centered around 
+the [Commander X16](https://www.commanderx16.com/) retro computer. Their [discord server](https://discord.gg/nS2PqEC) contains a small channel
+dedicated to Prog8. Other than that, use the issue tracker on github.
+
+
 Software license
 ----------------
 GNU GPL 3.0 (see file LICENSE), with exception for generated code:
@@ -59,6 +74,8 @@ What does Prog8 provide?
 - "c64": Commodore-64  (6502 like CPU)
 - "c128": Commodore-128  (6502 like CPU - the Z80 cpu mode is not supported)
 - "cx16": [CommanderX16](https://www.commanderx16.com)  (65c02 CPU)
+- "pet32": Commodore PET (experimental)
+- "atari": Atari 8 bit such as 800XL (experimental)
 - If you only use standard kernal and prog8 library routines, it is possible to compile the *exact same program* for different machines (just change the compiler target flag)
 
 

codeCore/build.gradle

@@ -26,7 +26,7 @@ compileTestKotlin {
 dependencies {
     // should have no dependencies to other modules
     implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8"
-    implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.17"
+    implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.18"
 }
 
 sourceSets {

codeCore/src/prog8/code/SymbolTable.kt

@@ -80,6 +80,16 @@ class SymbolTable(astProgram: PtProgram) : StNode(astProgram.name, StNodeType.GL
     }
 
     override fun lookup(scopedName: String) = flat[scopedName]
+
+    fun getLength(name: String): Int? {
+        val node = flat[name]
+        return when(node) {
+            is StMemVar -> node.length
+            is StMemorySlab -> node.size.toInt()
+            is StStaticVariable -> node.length
+            else -> null
+        }
+    }
 }
 
 
@@ -203,8 +213,7 @@ class StStaticVariable(name: String,
 
 
 class StConstant(name: String, val dt: DataType, val value: Double, astNode: PtNode) :
-    StNode(name, StNodeType.CONSTANT, astNode) {
-}
+    StNode(name, StNodeType.CONSTANT, astNode)
 
 
 class StMemVar(name: String,
@@ -226,12 +235,11 @@ class StMemorySlab(
     val align: UInt,
     astNode: PtNode
 ):
-    StNode(name, StNodeType.MEMORYSLAB, astNode) {
-}
+    StNode(name, StNodeType.MEMORYSLAB, astNode)
+
 
 class StSub(name: String, val parameters: List<StSubroutineParameter>, val returnType: DataType?, astNode: PtNode) :
-        StNode(name, StNodeType.SUBROUTINE, astNode) {
-}
+        StNode(name, StNodeType.SUBROUTINE, astNode)
 
 
 class StRomSub(name: String,

codeCore/src/prog8/code/SymbolTableMaker.kt

@@ -26,8 +26,6 @@ class SymbolTableMaker(private val program: PtProgram, private val options: Comp
                 PtMemMapped("P8ZP_SCRATCH_REG", DataType.UBYTE, options.compTarget.machine.zeropage.SCRATCH_REG, null, Position.DUMMY),
                 PtMemMapped("P8ZP_SCRATCH_W1", DataType.UWORD, options.compTarget.machine.zeropage.SCRATCH_W1, null, Position.DUMMY),
                 PtMemMapped("P8ZP_SCRATCH_W2", DataType.UWORD, options.compTarget.machine.zeropage.SCRATCH_W2, null, Position.DUMMY),
-                PtMemMapped("P8ESTACK_LO", DataType.ARRAY_UB, options.compTarget.machine.ESTACK_LO, 128u, Position.DUMMY),
-                PtMemMapped("P8ESTACK_HI", DataType.ARRAY_UB, options.compTarget.machine.ESTACK_HI, 128u, Position.DUMMY)
             ).forEach {
                 it.parent = program
                 st.add(StMemVar(it.name, it.type, it.address, it.arraySize?.toInt(), it))
@@ -94,6 +92,9 @@ class SymbolTableMaker(private val program: PtProgram, private val options: Comp
                     initialString = null
                     numElements = node.arraySize?.toInt()
                 }
+//                if(node.type in SplitWordArrayTypes) {
+//                    ... split array also add _lsb and _msb to symboltable?
+//                }
                 StStaticVariable(node.name, node.type, initialNumeric, initialString, initialArray, numElements, node.zeropage, node)
             }
             is PtBuiltinFunctionCall -> {

codeCore/src/prog8/code/ast/AstBase.kt

@@ -37,16 +37,17 @@ class PtNodeGroup : PtNode(Position.DUMMY)
 sealed class PtNamedNode(var name: String, position: Position): PtNode(position) {
     // Note that as an exception, the 'name' is not read-only
     // but a var. This is to allow for cheap node renames.
-    val scopedName: String by lazy {
-        var namedParent: PtNode = this.parent
-        if(namedParent is PtProgram)
-            name
-        else {
-            while (namedParent !is PtNamedNode)
-                namedParent = namedParent.parent
-            namedParent.scopedName + "." + name
+    val scopedName: String
+        get() {
+            var namedParent: PtNode = this.parent
+            return if(namedParent is PtProgram)
+                name
+            else {
+                while (namedParent !is PtNamedNode)
+                    namedParent = namedParent.parent
+                namedParent.scopedName + "." + name
+            }
         }
-    }
 }
 
 
@@ -63,7 +64,9 @@ class PtProgram(
         children.asSequence().filterIsInstance<PtBlock>()
 
     fun entrypoint(): PtSub? =
-        allBlocks().firstOrNull { it.name == "main" }?.children?.firstOrNull { it is PtSub && (it.name == "start" || it.name=="main.start") } as PtSub?
+        allBlocks().firstOrNull { it.name == "main" || it.name=="p8_main" }
+            ?.children
+            ?.firstOrNull { it is PtSub && (it.name == "start" || it.name=="main.start" || it.name=="p8_start" || it.name=="p8_main.p8_start") } as PtSub?
 }
 
 
@@ -71,6 +74,7 @@ class PtBlock(name: String,
               val address: UInt?,
               val library: Boolean,
               val forceOutput: Boolean,
+              val noSymbolPrefixing: Boolean,
               val alignment: BlockAlignment,
               val source: SourceCode,       // taken from the module the block is defined in.
               position: Position

codeCore/src/prog8/code/ast/AstExpressions.kt

@@ -118,6 +118,9 @@ class PtArrayIndexer(elementType: DataType, position: Position): PtExpression(el
     val splitWords: Boolean
         get() = variable.type in SplitWordArrayTypes
 
+    val usesPointerVariable: Boolean
+        get() = variable.type==DataType.UWORD
+
     init {
         require(elementType in NumericDatatypes)
     }

codeCore/src/prog8/code/ast/AstPrinter.kt

@@ -126,7 +126,6 @@ fun printAst(root: PtNode, skipLibraries: Boolean, output: (text: String) -> Uni
                 else
                     "->"
             }
-            else -> throw InternalCompilerException("unrecognised ast node $node")
         }
     }
 

codeCore/src/prog8/code/core/BuiltinFunctions.kt

@@ -78,14 +78,16 @@ val BuiltinFunctions: Map<String, FSignature> = mapOf(
     "reverse"   to FSignature(false, listOf(FParam("array", ArrayDatatypes)), null),
     // cmp returns a status in the carry flag, but not a proper return value
     "cmp"       to FSignature(false, listOf(FParam("value1", IntegerDatatypesNoBool), FParam("value2", NumericDatatypesNoBool)), null),
-    "prog8_lib_stringcompare"       to FSignature(true, listOf(FParam("str1", arrayOf(DataType.STR)), FParam("str2", arrayOf(DataType.STR))), DataType.BYTE),
+    "prog8_lib_stringcompare"     to FSignature(true, listOf(FParam("str1", arrayOf(DataType.STR)), FParam("str2", arrayOf(DataType.STR))), DataType.BYTE),
+    "prog8_lib_square_byte"       to FSignature(true, listOf(FParam("value", arrayOf(DataType.BYTE, DataType.UBYTE))), DataType.UBYTE),
+    "prog8_lib_square_word"       to FSignature(true, listOf(FParam("value", arrayOf(DataType.WORD, DataType.UWORD))), DataType.UWORD),
     "abs"       to FSignature(true, listOf(FParam("value", NumericDatatypesNoBool)), null),
     "abs__byte"  to FSignature(true, listOf(FParam("value", arrayOf(DataType.BYTE))), DataType.BYTE),
     "abs__word"  to FSignature(true, listOf(FParam("value", arrayOf(DataType.WORD))), DataType.WORD),
     "abs__float" to FSignature(true, listOf(FParam("value", arrayOf(DataType.FLOAT))), DataType.FLOAT),
     "len"       to FSignature(true, listOf(FParam("values", IterableDatatypes)), DataType.UWORD),
     // normal functions follow:
-    "sizeof"    to FSignature(true, listOf(FParam("object", DataType.values())), DataType.UBYTE),
+    "sizeof"    to FSignature(true, listOf(FParam("object", DataType.entries.toTypedArray())), DataType.UBYTE),
     "sgn"       to FSignature(true, listOf(FParam("value", NumericDatatypesNoBool)), DataType.BYTE),
     "sqrt"        to FSignature(true, listOf(FParam("value", NumericDatatypesNoBool)), null),
     "sqrt__ubyte" to FSignature(true, listOf(FParam("value", arrayOf(DataType.UBYTE))), DataType.UBYTE),
@@ -124,9 +126,7 @@ val BuiltinFunctions: Map<String, FSignature> = mapOf(
     "push"      to FSignature(false, listOf(FParam("value", ByteDatatypes)), null),
     "pushw"     to FSignature(false, listOf(FParam("value", WordDatatypes)), null),
     "rsave"     to FSignature(false, emptyList(), null),
-    "rsavex"    to FSignature(false, emptyList(), null),
     "rrestore"  to FSignature(false, emptyList(), null),
-    "rrestorex" to FSignature(false, emptyList(), null),
     "memory"    to FSignature(true, listOf(FParam("name", arrayOf(DataType.STR)), FParam("size", arrayOf(DataType.UWORD)), FParam("alignment", arrayOf(DataType.UWORD))), DataType.UWORD),
     "callfar"   to FSignature(false, listOf(FParam("bank", arrayOf(DataType.UBYTE)), FParam("address", arrayOf(DataType.UWORD)), FParam("arg", arrayOf(DataType.UWORD))), DataType.UWORD),
 )

codeCore/src/prog8/code/core/CompilationOptions.kt

@@ -13,16 +13,14 @@ class CompilationOptions(val output: OutputType,
                          val compTarget: ICompilationTarget,
                          // these are set later, based on command line arguments or options in the source code:
                          var loadAddress: UInt,
-                         var slowCodegenWarnings: Boolean = false,
+                         var warnSymbolShadowing: Boolean = false,
                          var optimize: Boolean = false,
-                         var optimizeFloatExpressions: Boolean = false,
                          var asmQuiet: Boolean = false,
                          var asmListfile: Boolean = false,
+                         var includeSourcelines: Boolean = false,
                          var experimentalCodegen: Boolean = false,
-                         var varsHigh: Boolean = false,
-                         var useNewExprCode: Boolean = false,
+                         var varsHighBank: Int? = null,
                          var splitWordArrays: Boolean = false,
-                         var evalStackBaseAddress: UInt? = null,
                          var outputDir: Path = Path(""),
                          var symbolDefs: Map<String, String> = emptyMap()
 ) {

codeCore/src/prog8/code/core/IErrorReporter.kt

@@ -3,6 +3,7 @@ package prog8.code.core
 interface IErrorReporter {
     fun err(msg: String, position: Position)
     fun warn(msg: String, position: Position)
+    fun undefined(symbol: List<String>, position: Position)
     fun noErrors(): Boolean
     fun report()
     fun finalizeNumErrors(numErrors: Int, numWarnings: Int) {

codeCore/src/prog8/code/core/IMachineDefinition.kt

@@ -13,8 +13,6 @@ interface IMachineDefinition {
     val FLOAT_MAX_NEGATIVE: Double
     val FLOAT_MAX_POSITIVE: Double
     val FLOAT_MEM_SIZE: Int
-    var ESTACK_LO: UInt
-    var ESTACK_HI: UInt
     val PROGRAM_LOAD_ADDRESS : UInt
     val BSSHIGHRAM_START: UInt
     val BSSHIGHRAM_END: UInt
@@ -29,11 +27,4 @@ interface IMachineDefinition {
     fun importLibs(compilerOptions: CompilationOptions, compilationTargetName: String): List<String>
     fun launchEmulator(selectedEmulator: Int, programNameWithPath: Path)
     fun isIOAddress(address: UInt): Boolean
-    fun overrideEvalStack(evalStackBaseAddress: UInt) {
-        require(evalStackBaseAddress and 255u == 0u)
-        ESTACK_LO = evalStackBaseAddress
-        ESTACK_HI = evalStackBaseAddress + 256u
-        require(ESTACK_LO !in golden.region && ESTACK_HI !in golden.region) { "user-set ESTACK can't be in GOLDEN ram" }
-    }
-
 }

codeCore/src/prog8/code/core/Position.kt

@@ -1,6 +1,6 @@
 package prog8.code.core
 
-import prog8.code.core.SourceCode.Companion.libraryFilePrefix
+import prog8.code.core.SourceCode.Companion.LIBRARYFILEPREFIX
 import java.nio.file.InvalidPathException
 import kotlin.io.path.Path
 import kotlin.io.path.absolute
@@ -10,7 +10,7 @@ data class Position(val file: String, val line: Int, val startCol: Int, val endC
     fun toClickableStr(): String {
         if(this===DUMMY)
             return ""
-        if(file.startsWith(libraryFilePrefix))
+        if(file.startsWith(LIBRARYFILEPREFIX))
             return "$file:$line:$startCol:"
         return try {
             val path = Path(file).absolute().normalize().toString()

codeCore/src/prog8/code/core/SourceCode.kt

@@ -54,12 +54,14 @@ sealed class SourceCode {
         /**
          * filename prefix to designate library files that will be retreived from internal resources rather than disk
          */
-        const val libraryFilePrefix = "library:"
-        const val stringSourcePrefix = "string:"
+        const val LIBRARYFILEPREFIX = "library:"
+        const val STRINGSOURCEPREFIX = "string:"
         val curdir: Path = Path(".").toAbsolutePath()
         fun relative(path: Path): Path = curdir.relativize(path.toAbsolutePath())
         fun isRegularFilesystemPath(pathString: String) =
-            !(pathString.startsWith(libraryFilePrefix) || pathString.startsWith(stringSourcePrefix))
+            !(pathString.startsWith(LIBRARYFILEPREFIX) || pathString.startsWith(STRINGSOURCEPREFIX))
+
+        fun isLibraryResource(path: String) = path.startsWith(LIBRARYFILEPREFIX)
     }
 
     /**
@@ -69,7 +71,7 @@ sealed class SourceCode {
     class Text(override val text: String): SourceCode() {
         override val isFromResources = false
         override val isFromFilesystem = false
-        override val origin = "$stringSourcePrefix${System.identityHashCode(text).toString(16)}"
+        override val origin = "$STRINGSOURCEPREFIX${System.identityHashCode(text).toString(16)}"
         override val name = "<unnamed-text>"
     }
 
@@ -110,7 +112,7 @@ sealed class SourceCode {
 
         override val isFromResources = true
         override val isFromFilesystem = false
-        override val origin = "$libraryFilePrefix$normalized"
+        override val origin = "$LIBRARYFILEPREFIX$normalized"
         override val text: String
         override val name: String
 
@@ -139,3 +141,33 @@ sealed class SourceCode {
         override val text: String = "<generated code node, no text representation>"
     }
 }
+
+
+object SourceLineCache {
+    private val cache = mutableMapOf<String, List<String>>()
+
+    private fun getCachedFile(file: String): List<String> {
+        val existing = cache[file]
+        if(existing!=null)
+            return existing
+        if (SourceCode.isRegularFilesystemPath(file)) {
+            val source = SourceCode.File(Path(file))
+            cache[file] = source.text.split('\n', '\r').map { it.trim() }
+            return cache.getValue(file)
+        } else if(file.startsWith(SourceCode.LIBRARYFILEPREFIX)) {
+            val source = SourceCode.Resource(file.drop(SourceCode.LIBRARYFILEPREFIX.length))
+            cache[file] = source.text.split('\n', '\r').map { it.trim()}
+            return cache.getValue(file)
+        }
+        return emptyList()
+    }
+
+    fun retrieveLine(position: Position): String? {
+        if (position.line>0) {
+            val lines = getCachedFile(position.file)
+            if(lines.isNotEmpty())
+                return lines[position.line-1]
+        }
+        return null
+    }
+}

codeCore/src/prog8/code/target/PETTarget.kt

@@ -0,0 +1,20 @@
+package prog8.code.target
+
+import prog8.code.core.Encoding
+import prog8.code.core.ICompilationTarget
+import prog8.code.core.IMemSizer
+import prog8.code.core.IStringEncoding
+import prog8.code.target.cbm.CbmMemorySizer
+import prog8.code.target.pet.PETMachineDefinition
+
+
+class PETTarget: ICompilationTarget, IStringEncoding by Encoder, IMemSizer by CbmMemorySizer {
+    override val name = NAME
+    override val machine = PETMachineDefinition()
+    override val supportedEncodings = setOf(Encoding.PETSCII, Encoding.SCREENCODES)
+    override val defaultEncoding = Encoding.PETSCII
+
+    companion object {
+        const val NAME = "pet32"
+    }
+}

codeCore/src/prog8/code/target/atari/AtariMachineDefinition.kt

@@ -13,10 +13,6 @@ class AtariMachineDefinition: IMachineDefinition {
     override val FLOAT_MEM_SIZE = 6
     override val PROGRAM_LOAD_ADDRESS = 0x2000u
 
-    // the 2*128 byte evaluation stack (1 page, on which bytes, words, and even floats are stored during calculations)
-    override var ESTACK_LO = 0x1b00u     //  $1b00-$1b7f inclusive      // TODO
-    override var ESTACK_HI = 0x1b80u     //  $1b80-$1bff inclusive      // TODO
-
     override val BSSHIGHRAM_START = 0u    // TODO
     override val BSSHIGHRAM_END = 0u      // TODO
 

codeCore/src/prog8/code/target/c128/C128MachineDefinition.kt

@@ -15,10 +15,6 @@ class C128MachineDefinition: IMachineDefinition {
     override val FLOAT_MEM_SIZE = Mflpt5.FLOAT_MEM_SIZE
     override val PROGRAM_LOAD_ADDRESS = 0x1c01u
 
-    // the 2*128 byte evaluation stack (1 page, on which bytes, words, and even floats are stored during calculations)
-    override var ESTACK_LO = 0x1b00u     //  $1b00-$1b7f inclusive
-    override var ESTACK_HI = 0x1b80u     //  $1b80-$1bff inclusive
-
     override val BSSHIGHRAM_START = 0u    // TODO
     override val BSSHIGHRAM_END = 0u      // TODO
 

codeCore/src/prog8/code/target/c64/C64MachineDefinition.kt

@@ -16,12 +16,8 @@ class C64MachineDefinition: IMachineDefinition {
     override val FLOAT_MEM_SIZE = Mflpt5.FLOAT_MEM_SIZE
     override val PROGRAM_LOAD_ADDRESS = 0x0801u
 
-    // the 2*128 byte evaluation stack (1 page, on which bytes, words, and even floats are stored during calculations)
-    override var ESTACK_LO = 0xcf00u     //  $cf00-$cf7f inclusive
-    override var ESTACK_HI = 0xcf80u     //  $cf80-$cfff inclusive
-
     override val BSSHIGHRAM_START = 0xc000u
-    override val BSSHIGHRAM_END = ESTACK_LO
+    override val BSSHIGHRAM_END = 0xd000u
 
     override lateinit var zeropage: Zeropage
     override lateinit var golden: GoldenRam
@@ -62,7 +58,7 @@ class C64MachineDefinition: IMachineDefinition {
 
     override fun initializeMemoryAreas(compilerOptions: CompilationOptions) {
         zeropage = C64Zeropage(compilerOptions)
-        golden = GoldenRam(compilerOptions, 0xc000u until ESTACK_LO)
+        golden = GoldenRam(compilerOptions, 0xc000u until 0xd000u)
     }
 
 }

codeCore/src/prog8/code/target/cx16/CX16MachineDefinition.kt

@@ -15,10 +15,6 @@ class CX16MachineDefinition: IMachineDefinition {
     override val FLOAT_MEM_SIZE = Mflpt5.FLOAT_MEM_SIZE
     override val PROGRAM_LOAD_ADDRESS = 0x0801u
 
-    // the 2*128 byte evaluation stack (1 page, on which bytes, words, and even floats are stored during calculations)
-    override var ESTACK_LO = 0x0700u        //  $0700-$077f inclusive
-    override var ESTACK_HI = 0x0780u        //  $0780-$07ff inclusive
-
     override val BSSHIGHRAM_START = 0xa000u     // hiram bank 1, 8Kb, assumed to be active
     override val BSSHIGHRAM_END = 0xc000u       // rom starts here.
 
@@ -64,7 +60,7 @@ class CX16MachineDefinition: IMachineDefinition {
 
     override fun initializeMemoryAreas(compilerOptions: CompilationOptions) {
         zeropage = CX16Zeropage(compilerOptions)
-        golden = GoldenRam(compilerOptions, 0x0400u until ESTACK_LO)
+        golden = GoldenRam(compilerOptions, 0x0400u until 0x0800u)
     }
 
 }

codeCore/src/prog8/code/target/cx16/CX16Zeropage.kt

@@ -40,7 +40,6 @@ class CX16Zeropage(options: CompilationOptions) : Zeropage(options) {
                 ZeropageType.DONTUSE -> {
                     free.clear() // don't use zeropage at all
                 }
-                else -> throw InternalCompilerException("for this machine target, zero page type 'floatsafe' is not available. ${options.zeropage}")
             }
 
             val distinctFree = free.distinct()

codeCore/src/prog8/code/target/pet/PETMachineDefinition.kt

@@ -0,0 +1,55 @@
+package prog8.code.target.pet
+
+import prog8.code.core.*
+import prog8.code.target.C64Target
+import prog8.code.target.cbm.Mflpt5
+import java.nio.file.Path
+
+
+class PETMachineDefinition: IMachineDefinition {
+
+    override val cpu = CpuType.CPU6502
+
+    override val FLOAT_MAX_POSITIVE = Mflpt5.FLOAT_MAX_POSITIVE
+    override val FLOAT_MAX_NEGATIVE = Mflpt5.FLOAT_MAX_NEGATIVE
+    override val FLOAT_MEM_SIZE = Mflpt5.FLOAT_MEM_SIZE
+    override val PROGRAM_LOAD_ADDRESS = 0x0401u
+
+    override val BSSHIGHRAM_START = 0xffffu
+    override val BSSHIGHRAM_END = 0xffffu
+
+    override lateinit var zeropage: Zeropage
+    override lateinit var golden: GoldenRam
+
+    override fun getFloatAsmBytes(num: Number) = Mflpt5.fromNumber(num).makeFloatFillAsm()
+
+    override fun importLibs(compilerOptions: CompilationOptions, compilationTargetName: String): List<String> {
+        return if (compilerOptions.launcher == CbmPrgLauncherType.BASIC || compilerOptions.output == OutputType.PRG)
+            listOf("syslib")
+        else
+            emptyList()
+    }
+
+    override fun launchEmulator(selectedEmulator: Int, programNameWithPath: Path) {
+        if(selectedEmulator!=1) {
+            System.err.println("The pet target only supports the main emulator (Vice).")
+            return
+        }
+
+        println("\nStarting PET emulator...")
+        val viceMonlist = C64Target.viceMonListName(programNameWithPath.toString())
+        val cmdline = listOf("xpet", "-model", "4032", "-ramsize", "32", "-videosize", "40", "-silent", "-moncommands", viceMonlist,
+                "-autostartprgmode", "1", "-autostart-warp", "-autostart", "${programNameWithPath}.prg")
+        val processb = ProcessBuilder(cmdline).inheritIO()
+        val process=processb.start()
+        process.waitFor()
+    }
+
+    override fun isIOAddress(address: UInt): Boolean = address in 0xe800u..0xe8ffu
+
+    override fun initializeMemoryAreas(compilerOptions: CompilationOptions) {
+        zeropage = PETZeropage(compilerOptions)
+        // there's no golden ram.
+    }
+
+}

codeCore/src/prog8/code/target/pet/PETZeropage.kt

@@ -0,0 +1,58 @@
+package prog8.code.target.pet
+
+import prog8.code.core.CompilationOptions
+import prog8.code.core.InternalCompilerException
+import prog8.code.core.Zeropage
+import prog8.code.core.ZeropageType
+
+
+// reference: http://www.zimmers.net/cbmpics/cbm/PETx/petmem.txt
+
+class PETZeropage(options: CompilationOptions) : Zeropage(options) {
+
+    override val SCRATCH_B1 = 0xb3u      // temp storage for a single byte
+    override val SCRATCH_REG = 0xb4u     // temp storage for a register, must be B1+1
+    override val SCRATCH_W1 = 0xb6u      // temp storage 1 for a word
+    override val SCRATCH_W2 = 0xb8u      // temp storage 2 for a word
+
+    init {
+        if (options.floats) {
+            throw InternalCompilerException("PET target doesn't yet support floating point routines")
+        }
+
+        if (options.floats && options.zeropage !in arrayOf(
+                ZeropageType.FLOATSAFE,
+                ZeropageType.BASICSAFE,
+                ZeropageType.DONTUSE
+            ))
+            throw InternalCompilerException("when floats are enabled, zero page type should be 'floatsafe' or 'basicsafe' or 'dontuse'")
+
+        when (options.zeropage) {
+            ZeropageType.FULL -> {
+                free.addAll(0x00u..0xffu)
+                free.removeAll(listOf(0x8du, 0x8eu, 0x8fu, 0x97u, 0x98u, 0x99u, 0x9au, 0x9bu, 0x9eu, 0xa7u, 0xa8u, 0xa9u, 0xaau))        // these are updated/used by IRQ
+            }
+            ZeropageType.KERNALSAFE -> {
+                free.addAll(0x00u..0xffu)
+                free.removeAll(listOf(0x8du, 0x8eu, 0x8fu, 0x97u, 0x98u, 0x99u, 0x9au, 0x9bu, 0x9eu, 0xa7u, 0xa8u, 0xa9u, 0xaau))        // these are updated/used by IRQ
+            }
+            ZeropageType.FLOATSAFE,
+            ZeropageType.BASICSAFE -> {
+                free.addAll(0xb3u..0xbau)       // TODO more?
+            }
+            ZeropageType.DONTUSE -> {
+                free.clear()  // don't use zeropage at all
+            }
+        }
+
+        val distinctFree = free.distinct()
+        free.clear()
+        free.addAll(distinctFree)
+
+        removeReservedFromFreePool()
+    }
+
+    override fun allocateCx16VirtualRegisters() {
+        TODO("Not known if PET can put the virtual regs in ZP")
+    }
+}

codeCore/src/prog8/code/target/virtual/VirtualMachineDefinition.kt

@@ -15,8 +15,6 @@ class VirtualMachineDefinition: IMachineDefinition {
     override val FLOAT_MEM_SIZE = 4             // 32-bits floating point
     override val PROGRAM_LOAD_ADDRESS = 0u      // not actually used
 
-    override var ESTACK_LO = 0u                 // not actually used
-    override var ESTACK_HI = 0u                 // not actually used
     override val BSSHIGHRAM_START = 0u          // not actually used
     override val BSSHIGHRAM_END = 0u            // not actually used
     override lateinit var zeropage: Zeropage    // not actually used

codeGenCpu6502/build.gradle

@@ -28,7 +28,7 @@ dependencies {
     implementation project(':codeCore')
     implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8"
     // implementation "org.jetbrains.kotlin:kotlin-reflect"
-    implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.17"
+    implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.18"
 
     testImplementation 'io.kotest:kotest-runner-junit5-jvm:5.6.2'
 }

codeGenCpu6502/src/prog8/codegen/cpu6502/AsmGen.kt

@@ -3,6 +3,7 @@ package prog8.codegen.cpu6502
 import com.github.michaelbull.result.fold
 import prog8.code.StNodeType
 import prog8.code.SymbolTable
+import prog8.code.SymbolTableMaker
 import prog8.code.ast.*
 import prog8.code.core.*
 import prog8.codegen.cpu6502.assignment.*
@@ -14,18 +15,178 @@ import kotlin.io.path.writeLines
 internal const val subroutineFloatEvalResultVar1 = "prog8_float_eval_result1"
 internal const val subroutineFloatEvalResultVar2 = "prog8_float_eval_result2"
 
-class AsmGen6502: ICodeGeneratorBackend {
+class AsmGen6502(val prefixSymbols: Boolean): ICodeGeneratorBackend {
     override fun generate(
         program: PtProgram,
         symbolTable: SymbolTable,
         options: CompilationOptions,
         errors: IErrorReporter
     ): IAssemblyProgram? {
-        val asmgen = AsmGen6502Internal(program, symbolTable, options, errors)
+        val st = if(prefixSymbols) prefixSymbols(program, options, symbolTable) else symbolTable
+        val asmgen = AsmGen6502Internal(program, st, options, errors)
         return asmgen.compileToAssembly()
     }
+
+    private fun prefixSymbols(program: PtProgram, options: CompilationOptions, st: SymbolTable): SymbolTable {
+        val nodesToPrefix = mutableListOf<Pair<PtNode, Int>>()              // parent + index
+        val functionCallsToPrefix = mutableListOf<Pair<PtNode, Int>>()      // parent + index
+
+        fun prefixNamedNode(node: PtNamedNode) {
+            node.name = "p8_${node.name}"
+        }
+
+        fun prefixSymbols(node: PtNode) {
+            when(node) {
+                is PtAsmSub -> {
+                    prefixNamedNode(node)
+                    node.parameters.forEach { (_, param) -> prefixNamedNode(param) }
+                }
+                is PtSub -> {
+                    prefixNamedNode(node)
+                    node.parameters.forEach { prefixNamedNode(it) }
+                }
+                is PtFunctionCall -> {
+                    val stNode = st.lookup(node.name)!!
+                    if(stNode.astNode.definingBlock()?.noSymbolPrefixing!=true) {
+                        val index = node.parent.children.indexOf(node)
+                        functionCallsToPrefix += node.parent to index
+                    }
+                }
+                is PtIdentifier -> {
+                    var lookupName = node.name
+                    if(node.type in SplitWordArrayTypes && (lookupName.endsWith("_lsb") || lookupName.endsWith("_msb"))) {
+                        lookupName = lookupName.dropLast(4)
+                    }
+                    val stNode = st.lookup(lookupName)!!
+                    if(stNode.astNode.definingBlock()?.noSymbolPrefixing!=true) {
+                        val index = node.parent.children.indexOf(node)
+                        nodesToPrefix += node.parent to index
+                    }
+                }
+                is PtJump -> {
+                    if(node.identifier!=null) {
+                        val stNode = st.lookup(node.identifier!!.name)!!
+                        if(stNode.astNode.definingBlock()?.noSymbolPrefixing!=true) {
+                            val index = node.parent.children.indexOf(node)
+                            nodesToPrefix += node.parent to index
+                        }
+                    }
+                    else if(node.generatedLabel!=null) {
+                        val stNode = st.lookup(node.generatedLabel!!)!!
+                        if(stNode.astNode.definingBlock()?.noSymbolPrefixing!=true) {
+                            val index = node.parent.children.indexOf(node)
+                            nodesToPrefix += node.parent to index
+                        }
+                    }
+                }
+                is PtBlock -> prefixNamedNode(node)
+                is PtConstant -> prefixNamedNode(node)
+                is PtLabel -> prefixNamedNode(node)
+                is PtMemMapped -> prefixNamedNode(node)
+                is PtSubroutineParameter -> prefixNamedNode(node)
+                is PtVariable -> {
+                    val index = node.parent.children.indexOf(node)
+                    nodesToPrefix += node.parent to index
+                }
+                else -> { }
+            }
+            node.children.forEach { prefixSymbols(it) }
+        }
+
+        program.allBlocks().forEach { block ->
+            if (!block.noSymbolPrefixing) {
+                prefixSymbols(block)
+            }
+        }
+
+        nodesToPrefix.forEach { (parent, index) ->
+            val node = parent.children[index]
+            when(node) {
+                is PtIdentifier -> parent.children[index] = node.prefix(parent, st)
+                is PtFunctionCall ->  throw AssemblyError("PtFunctionCall should be processed in their own list, last")
+                is PtJump -> parent.children[index] = node.prefix(parent, st)
+                is PtVariable -> parent.children[index] = node.prefix(st)
+                else -> throw AssemblyError("weird node to prefix $node")
+            }
+        }
+
+        // reversed so inner calls (such as arguments to a function call) get processed before the actual function call itself
+        functionCallsToPrefix.reversed().forEach { (parent, index) ->
+            val node = parent.children[index]
+            if(node is PtFunctionCall) {
+                parent.children[index] = node.prefix(parent)
+            } else {
+                throw AssemblyError("expected PtFunctionCall")
+            }
+        }
+
+        return SymbolTableMaker(program, options).make()
+    }
 }
 
+private fun PtVariable.prefix(st: SymbolTable): PtVariable {
+    name = name.split('.').map {"p8_$it" }.joinToString(".")
+    if(value==null)
+        return this
+
+    val arrayValue = value as? PtArray
+    return if(arrayValue!=null && arrayValue.children.any { it !is PtNumber} ) {
+        val newValue = PtArray(arrayValue.type, arrayValue.position)
+        arrayValue.children.forEach { elt ->
+            when(elt) {
+                is PtIdentifier -> newValue.add(elt.prefix(arrayValue, st))
+                is PtNumber -> newValue.add(elt)
+                is PtAddressOf -> {
+                    if(elt.definingBlock()?.noSymbolPrefixing==true)
+                        newValue.add(elt)
+                    else {
+                        val newAddr = PtAddressOf(elt.position)
+                        newAddr.children.add(elt.identifier.prefix(newAddr, st))
+                        newAddr.parent = arrayValue
+                        newValue.add(newAddr)
+                    }
+                }
+                else -> throw AssemblyError("weird array value element $elt")
+            }
+        }
+        PtVariable(name, type, zeropage, newValue, arraySize, position)
+    }
+    else this
+}
+
+private fun PtJump.prefix(parent: PtNode, st: SymbolTable): PtJump {
+    val jump = if(identifier!=null) {
+        val prefixedIdent = identifier!!.prefix(this, st)
+        PtJump(prefixedIdent, address, generatedLabel, position)
+    } else {
+        val prefixedLabel = generatedLabel!!.split('.').map {"p8_$it" }.joinToString(".")
+        PtJump(null, address, prefixedLabel, position)
+    }
+    jump.parent = parent
+    return jump
+}
+
+private fun PtFunctionCall.prefix(parent: PtNode): PtFunctionCall {
+    val newName = name.split('.').map {"p8_$it" }.joinToString(".")
+    val call = PtFunctionCall(newName, void, type, position)
+    call.children.addAll(children)
+    call.children.forEach { it.parent = call }
+    call.parent = parent
+    return call
+}
+
+private fun PtIdentifier.prefix(parent: PtNode, st: SymbolTable): PtIdentifier {
+    val target = st.lookup(name)
+    if(target?.astNode?.definingBlock()?.noSymbolPrefixing==true)
+        return this
+
+    val newName = name.split('.').map { "p8_$it" }.joinToString(".")
+    val node = PtIdentifier(newName, type, position)
+    node.parent = parent
+    return node
+}
+
+
 class AsmGen6502Internal (
     val program: PtProgram,
     internal val symbolTable: SymbolTable,
@@ -40,12 +201,12 @@ class AsmGen6502Internal (
     private val allocator = VariableAllocator(symbolTable, options, errors)
     private val assemblyLines = mutableListOf<String>()
     private val breakpointLabels = mutableListOf<String>()
-    private val forloopsAsmGen = ForLoopsAsmGen(program, this, zeropage)
+    private val forloopsAsmGen = ForLoopsAsmGen(this, zeropage)
     private val postincrdecrAsmGen = PostIncrDecrAsmGen(program, this)
     private val functioncallAsmGen = FunctionCallAsmGen(program, this)
     private val programGen = ProgramAndVarsGen(program, options, errors, symbolTable, functioncallAsmGen, this, allocator, zeropage)
-    private val assignmentAsmGen = AssignmentAsmGen(program, symbolTable, this, allocator)
-    private val expressionsAsmGen = ExpressionsAsmGen(program, this, allocator)
+    private val anyExprGen = AnyExprAsmGen(this)
+    private val assignmentAsmGen = AssignmentAsmGen(program, this, anyExprGen, allocator)
     private val builtinFunctionsAsmGen = BuiltinFunctionsAsmGen(program, this, assignmentAsmGen)
 
     fun compileToAssembly(): IAssemblyProgram? {
@@ -77,8 +238,19 @@ class AsmGen6502Internal (
 
     internal fun isTargetCpu(cpu: CpuType) = options.compTarget.machine.cpu == cpu
 
+    private var lastSourceLineNumber: Int = -1
+
     internal fun outputSourceLine(node: PtNode) {
-        out(" ;\tsrc line: ${node.position.file}:${node.position.line}")
+        if(!options.includeSourcelines || node.position===Position.DUMMY || node.position.line==lastSourceLineNumber)
+            return
+
+        lastSourceLineNumber = node.position.line
+        val srcComment = "\t; source: ${node.position.file}:${node.position.line}"
+        val line = SourceLineCache.retrieveLine(node.position)
+        if(line==null)
+            out(srcComment, false)
+        else
+            out("$srcComment   $line", false)
     }
 
     internal fun out(str: String, splitlines: Boolean = true) {
@@ -231,32 +403,6 @@ class AsmGen6502Internal (
         return name2.replace("prog8_lib.P8ZP_SCRATCH_", "P8ZP_SCRATCH_")    // take care of the 'hooks' to the temp vars -> reference zp symbols directly
     }
 
-    internal fun saveRegisterLocal(register: CpuRegister, scope: IPtSubroutine) {
-        if (isTargetCpu(CpuType.CPU65c02)) {
-            // just use the cpu's stack for all registers, shorter code
-            when (register) {
-                CpuRegister.A -> out("  pha")
-                CpuRegister.X -> out("  phx")
-                CpuRegister.Y -> out("  phy")
-            }
-        } else {
-            when (register) {
-                CpuRegister.A -> {
-                    // just use the stack, only for A
-                    out("  pha")
-                }
-                CpuRegister.X -> {
-                    out("  stx  prog8_regsaveX")
-                    subroutineExtra(scope).usedRegsaveX = true
-                }
-                CpuRegister.Y -> {
-                    out("  sty  prog8_regsaveY")
-                    subroutineExtra(scope).usedRegsaveY = true
-                }
-            }
-        }
-    }
-
     internal fun saveRegisterStack(register: CpuRegister, keepA: Boolean) {
         when (register) {
             CpuRegister.A -> out("  pha")
@@ -281,24 +427,6 @@ class AsmGen6502Internal (
         }
     }
 
-    internal fun restoreRegisterLocal(register: CpuRegister) {
-        if (isTargetCpu(CpuType.CPU65c02)) {
-            when (register) {
-                // this just used the stack, for all registers. Shorter code.
-                CpuRegister.A -> out("  pla")
-                CpuRegister.X -> out("  plx")
-                CpuRegister.Y -> out("  ply")
-            }
-
-        } else {
-            when (register) {
-                CpuRegister.A -> out("  pla")   // this just used the stack but only for A
-                CpuRegister.X -> out("  ldx  prog8_regsaveX")
-                CpuRegister.Y -> out("  ldy  prog8_regsaveY")
-            }
-        }
-    }
-
     internal fun restoreRegisterStack(register: CpuRegister, keepA: Boolean) {
         when (register) {
             CpuRegister.A -> {
@@ -418,22 +546,12 @@ class AsmGen6502Internal (
         }
     }
 
-    @Deprecated("avoid calling this as it generates slow evalstack based code")
-    internal fun translateExpression(expression: PtExpression) =
-            expressionsAsmGen.translateExpression(expression)
-
-    internal fun translateBuiltinFunctionCallExpression(bfc: PtBuiltinFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?): DataType? =
-            builtinFunctionsAsmGen.translateFunctioncallExpression(bfc, resultToStack, resultRegister)
+    internal fun translateBuiltinFunctionCallExpression(bfc: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?): DataType? =
+            builtinFunctionsAsmGen.translateFunctioncallExpression(bfc, resultRegister)
 
     internal fun translateFunctionCall(functionCallExpr: PtFunctionCall) =
             functioncallAsmGen.translateFunctionCall(functionCallExpr)
 
-    internal fun saveXbeforeCall(functionCall: PtFunctionCall)  =
-            functioncallAsmGen.saveXbeforeCall(functionCall)
-
-    internal fun restoreXafterCall(functionCall: PtFunctionCall) =
-            functioncallAsmGen.restoreXafterCall(functionCall)
-
     internal fun translateNormalAssignment(assign: AsmAssignment, scope: IPtSubroutine?) =
             assignmentAsmGen.translateNormalAssignment(assign, scope)
 
@@ -462,7 +580,6 @@ class AsmGen6502Internal (
     }
 
     internal fun assignExpressionTo(value: PtExpression, target: AsmAssignTarget) {
-        // don't use translateExpression() to avoid evalstack
         when (target.datatype) {
             in ByteDatatypes -> {
                 assignExpressionToRegister(value, RegisterOrPair.A)
@@ -513,7 +630,6 @@ class AsmGen6502Internal (
     private fun translate(stmt: PtIfElse) {
         val condition = stmt.condition as? PtBinaryExpression
         if(condition!=null) {
-            require(!options.useNewExprCode)
             requireComparisonExpression(condition)  // IfStatement: condition must be of form  'x <comparison> <value>'
             if (stmt.elseScope.children.isEmpty()) {
                 val jump = stmt.ifScope.children.singleOrNull()
@@ -641,65 +757,48 @@ class AsmGen6502Internal (
     private fun repeatWordCount(count: Int, stmt: PtRepeatLoop) {
         require(count in 257..65535) { "invalid repeat count ${stmt.position}" }
         val repeatLabel = makeLabel("repeat")
-        if(isTargetCpu(CpuType.CPU65c02)) {
-            val counterVar = createRepeatCounterVar(DataType.UWORD, true, stmt)
-            out("""
-                lda  #<$count
-                ldy  #>$count
-                sta  $counterVar
-                sty  $counterVar+1
+        val counterVar = createRepeatCounterVar(DataType.UWORD, isTargetCpu(CpuType.CPU65c02), stmt)
+        // the iny + double dec is microoptimization of the 16 bit loop
+        out("""
+            ldy  #>$count
+            lda  #<$count
+            beq  +
+            iny
++           sta  $counterVar
+            sty  $counterVar+1
 $repeatLabel""")
-            translate(stmt.statements)
-            out("""
-                lda  $counterVar
-                bne  +
-                dec  $counterVar+1
-+               dec  $counterVar
-                lda  $counterVar
-                ora  $counterVar+1
-                bne  $repeatLabel""")
-        } else {
-            val counterVar = createRepeatCounterVar(DataType.UWORD, false, stmt)
-            out("""
-                lda  #<$count
-                ldy  #>$count
-                sta  $counterVar
-                sty  $counterVar+1
-$repeatLabel""")
-            translate(stmt.statements)
-            out("""
-                lda  $counterVar
-                bne  +
-                dec  $counterVar+1
-+               dec  $counterVar
-                lda  $counterVar
-                ora  $counterVar+1
-                bne  $repeatLabel""")
-        }
+        translate(stmt.statements)
+        out("""
+            dec  $counterVar
+            bne  $repeatLabel
+            dec  $counterVar+1
+            bne  $repeatLabel""")
     }
 
     private fun repeatWordCountInAY(endLabel: String, stmt: PtRepeatLoop) {
         // note: A/Y must have been loaded with the number of iterations!
-        // no need to explicitly test for 0 iterations as this is done in the countdown logic below
+        // the iny + double dec is microoptimization of the 16 bit loop
         val repeatLabel = makeLabel("repeat")
         val counterVar = createRepeatCounterVar(DataType.UWORD, false, stmt)
         out("""
-                sta  $counterVar
-                sty  $counterVar+1
-$repeatLabel    lda  $counterVar
-                bne  +
-                lda  $counterVar+1
-                beq  $endLabel
-                lda  $counterVar
-                bne  +
-                dec  $counterVar+1
-+               dec  $counterVar
-""")
+            cmp  #0
+            beq  +
+            iny
++           sta  $counterVar
+            sty  $counterVar+1
+            ora  $counterVar+1
+            beq  $endLabel
+$repeatLabel""")
         translate(stmt.statements)
-        jmp(repeatLabel)
+        out("""
+            dec  $counterVar
+            bne  $repeatLabel
+            dec  $counterVar+1
+            bne  $repeatLabel""")
         out(endLabel)
     }
 
+
     private fun repeatByteCount(count: Int, stmt: PtRepeatLoop) {
         require(count in 2..256) { "invalid repeat count ${stmt.position}" }
         val repeatLabel = makeLabel("repeat")
@@ -762,7 +861,7 @@ $repeatLabel    lda  $counterVar
             DataType.UBYTE, DataType.UWORD -> {
                 val result = zeropage.allocate(counterVar, dt, null, stmt.position, errors)
                 result.fold(
-                    success = { (address, _) -> asmInfo.extraVars.add(Triple(dt, counterVar, address)) },
+                    success = { (address, _, _) -> asmInfo.extraVars.add(Triple(dt, counterVar, address)) },
                     failure = { asmInfo.extraVars.add(Triple(dt, counterVar, null)) }  // allocate normally
                 )
                 return counterVar
@@ -902,7 +1001,10 @@ $repeatLabel    lda  $counterVar
     }
 
     private fun translate(asm: PtInlineAssembly) {
-        assemblyLines.add(asm.assembly.trimEnd().trimStart('\r', '\n'))
+        if(asm.isIR)
+            throw AssemblyError("%asm containing IR code cannot be translated to 6502 assembly")
+        else
+            assemblyLines.add(asm.assembly.trimEnd().trimStart('\r', '\n'))
     }
 
     private fun translate(incbin: PtIncludeBinary) {
@@ -941,20 +1043,6 @@ $repeatLabel    lda  $counterVar
         }
     }
 
-    internal fun signExtendStackLsb(valueDt: DataType) {
-        // sign extend signed byte on stack to signed word on stack
-        when(valueDt) {
-            DataType.UBYTE -> {
-                if(isTargetCpu(CpuType.CPU65c02))
-                    out("  stz  P8ESTACK_HI+1,x")
-                else
-                    out("  lda  #0 |  sta  P8ESTACK_HI+1,x")
-            }
-            DataType.BYTE -> out("  jsr  prog8_lib.sign_extend_stack_byte")
-            else -> throw AssemblyError("need byte type")
-        }
-    }
-
     internal fun signExtendVariableLsb(asmvar: String, valueDt: DataType) {
         // sign extend signed byte in a var to a full word in that variable
         when(valueDt) {
@@ -990,18 +1078,10 @@ $repeatLabel    lda  $counterVar
     }
 
     internal fun pointerViaIndexRegisterPossible(pointerOffsetExpr: PtExpression): Pair<PtExpression, PtExpression>? {
-        val left: PtExpression
-        val right: PtExpression
-        val operator: String
-
-        if (pointerOffsetExpr is PtBinaryExpression) {
-            require(!options.useNewExprCode)
-            operator = pointerOffsetExpr.operator
-            left = pointerOffsetExpr.left
-            right = pointerOffsetExpr.right
-        }
-        else return null
-
+        if (pointerOffsetExpr !is PtBinaryExpression) return null
+        val operator = pointerOffsetExpr.operator
+        val left = pointerOffsetExpr.left
+        val right = pointerOffsetExpr.right
         if (operator != "+") return null
         val leftDt = left.type
         val rightDt = right.type
@@ -1113,10 +1193,7 @@ $repeatLabel    lda  $counterVar
     }
 
     internal fun findSubroutineParameter(name: String, asmgen: AsmGen6502Internal): PtSubroutineParameter? {
-        val stScope = asmgen.symbolTable.lookup(name)
-        require(stScope!=null) {
-            "invalid name lookup $name"
-        }
+        val stScope = asmgen.symbolTable.lookup(name) ?: return null
         val node = stScope.astNode
         if(node is PtSubroutineParameter)
             return node
@@ -1594,7 +1671,7 @@ $repeatLabel    lda  $counterVar
                 }
                 else if (left is PtMemoryByte) {
                     return if(rightConstVal.number.toInt()!=0) {
-                        translateDirectMemReadExpressionToRegAorStack(left, false)
+                        translateDirectMemReadExpressionToRegA(left)
                         code("#${rightConstVal.number.toInt()}")
                     }
                     else
@@ -1741,7 +1818,7 @@ $repeatLabel    lda  $counterVar
                         out("  beq $jumpIfFalseLabel")
                 }
                 else if (left is PtMemoryByte) {
-                    translateDirectMemReadExpressionToRegAorStack(left, false)
+                    translateDirectMemReadExpressionToRegA(left)
                     return if(rightConstVal.number.toInt()!=0)
                         code("#${rightConstVal.number.toInt()}")
                     else
@@ -1902,7 +1979,7 @@ $repeatLabel    lda  $counterVar
                         out("  bne  $jumpIfFalseLabel")
                 }
                 else if (left is PtMemoryByte) {
-                    translateDirectMemReadExpressionToRegAorStack(left, false)
+                    translateDirectMemReadExpressionToRegA(left)
                     return if(rightConstVal.number.toInt()!=0)
                         code("#${rightConstVal.number.toInt()}")
                     else
@@ -2066,7 +2143,7 @@ $repeatLabel    lda  $counterVar
                 }
                 else if (left is PtMemoryByte) {
                     if(rightConstVal.number.toInt()!=0) {
-                        translateDirectMemReadExpressionToRegAorStack(left, false)
+                        translateDirectMemReadExpressionToRegA(left)
                         code("#${rightConstVal.number.toInt()}")
                     }
                     return
@@ -2229,7 +2306,7 @@ $repeatLabel    lda  $counterVar
                         out("  bne  $jumpIfFalseLabel")
                 }
                 else if (left is PtMemoryByte) {
-                    translateDirectMemReadExpressionToRegAorStack(left, false)
+                    translateDirectMemReadExpressionToRegA(left)
                     return if(rightConstVal.number.toInt()!=0)
                         code("#${rightConstVal.number.toInt()}")
                     else
@@ -2275,7 +2352,7 @@ $repeatLabel    lda  $counterVar
                         out("  beq  $jumpIfFalseLabel")
                 }
                 else if (left is PtMemoryByte) {
-                    translateDirectMemReadExpressionToRegAorStack(left, false)
+                    translateDirectMemReadExpressionToRegA(left)
                     return if(rightConstVal.number.toInt()!=0)
                         code("#${rightConstVal.number.toInt()}")
                     else
@@ -2731,22 +2808,14 @@ $repeatLabel    lda  $counterVar
 +""")
     }
 
-    internal fun translateDirectMemReadExpressionToRegAorStack(expr: PtMemoryByte, pushResultOnEstack: Boolean) {
+    internal fun translateDirectMemReadExpressionToRegA(expr: PtMemoryByte) {
 
         fun assignViaExprEval() {
             assignExpressionToVariable(expr.address, "P8ZP_SCRATCH_W2", DataType.UWORD)
             if (isTargetCpu(CpuType.CPU65c02)) {
-                if (pushResultOnEstack) {
-                    out("  lda  (P8ZP_SCRATCH_W2) |  dex |  sta  P8ESTACK_LO+1,x")
-                } else {
-                    out("  lda  (P8ZP_SCRATCH_W2)")
-                }
+                out("  lda  (P8ZP_SCRATCH_W2)")
             } else {
-                if (pushResultOnEstack) {
-                    out("  ldy  #0 |  lda  (P8ZP_SCRATCH_W2),y |  dex |  sta  P8ESTACK_LO+1,x")
-                } else {
-                    out("  ldy  #0 |  lda  (P8ZP_SCRATCH_W2),y")
-                }
+                out("  ldy  #0 |  lda  (P8ZP_SCRATCH_W2),y")
             }
         }
 
@@ -2754,22 +2823,14 @@ $repeatLabel    lda  $counterVar
             is PtNumber -> {
                 val address = (expr.address as PtNumber).number.toInt()
                 out("  lda  ${address.toHex()}")
-                if(pushResultOnEstack)
-                    out("  sta  P8ESTACK_LO,x |  dex")
             }
             is PtIdentifier -> {
                 // the identifier is a pointer variable, so read the value from the address in it
                 loadByteFromPointerIntoA(expr.address as PtIdentifier)
-                if(pushResultOnEstack)
-                    out("  sta  P8ESTACK_LO,x |  dex")
             }
             is PtBinaryExpression -> {
-                require(!options.useNewExprCode)
                 val addrExpr = expr.address as PtBinaryExpression
-                if(tryOptimizedPointerAccessWithA(addrExpr, addrExpr.operator, false)) {
-                    if(pushResultOnEstack)
-                        out("  sta  P8ESTACK_LO,x |  dex")
-                } else {
+                if(!tryOptimizedPointerAccessWithA(addrExpr, addrExpr.operator, false)) {
                     assignViaExprEval()
                 }
             }
@@ -2965,6 +3026,18 @@ $repeatLabel    lda  $counterVar
         }
     }
 
+    internal fun pushFAC1() {
+        out("  jsr  floats.pushFAC1")
+    }
+
+    internal fun popFAC1() {
+        out("  clc |  jsr  floats.popFAC")
+    }
+
+    internal fun popFAC2() {
+        out("  sec |  jsr  floats.popFAC")
+    }
+
     internal fun needAsaveForExpr(arg: PtExpression): Boolean =
         arg !is PtNumber && arg !is PtIdentifier && (arg !is PtMemoryByte || !arg.isSimple())
 
@@ -2997,9 +3070,6 @@ $repeatLabel    lda  $counterVar
  * it's more consistent to only define these attributes on a Subroutine node.
  */
 internal class SubroutineExtraAsmInfo {
-    var usedRegsaveA = false
-    var usedRegsaveX = false
-    var usedRegsaveY = false
     var usedFloatEvalResultVar1 = false
     var usedFloatEvalResultVar2 = false
 

codeGenCpu6502/src/prog8/codegen/cpu6502/AsmOptimizer.kt

@@ -15,21 +15,7 @@ internal fun optimizeAssembly(lines: MutableList<String>, machine: IMachineDefin
 
     var linesByFour = getLinesBy(lines, 4)
 
-    var mods = optimizeUselessStackByteWrites(linesByFour)
-    if(mods.isNotEmpty()) {
-        apply(mods, lines)
-        linesByFour = getLinesBy(lines, 4)
-        numberOfOptimizations++
-    }
-
-    mods = optimizeIncDec(linesByFour)
-    if(mods.isNotEmpty()) {
-        apply(mods, lines)
-        linesByFour = getLinesBy(lines, 4)
-        numberOfOptimizations++
-    }
-
-    mods = optimizeCmpSequence(linesByFour)
+    var mods = optimizeIncDec(linesByFour)
     if(mods.isNotEmpty()) {
         apply(mods, lines)
         linesByFour = getLinesBy(lines, 4)
@@ -43,7 +29,21 @@ internal fun optimizeAssembly(lines: MutableList<String>, machine: IMachineDefin
         numberOfOptimizations++
     }
 
-    mods= optimizeJsrRtsAndOtherCombinations(linesByFour)
+    mods = optimizeJsrRtsAndOtherCombinations(linesByFour)
+    if(mods.isNotEmpty()) {
+        apply(mods, lines)
+        linesByFour = getLinesBy(lines, 4)
+        numberOfOptimizations++
+    }
+
+    mods = optimizeUselessPushPopStack(linesByFour)
+    if(mods.isNotEmpty()) {
+        apply(mods, lines)
+        linesByFour = getLinesBy(lines, 4)
+        numberOfOptimizations++
+    }
+
+    mods = optimizeUnneededTempvarInAdd(linesByFour)
     if(mods.isNotEmpty()) {
         apply(mods, lines)
         linesByFour = getLinesBy(lines, 4)
@@ -90,44 +90,6 @@ private fun getLinesBy(lines: MutableList<String>, windowSize: Int) =
 // all lines (that aren't empty or comments) in sliding windows of certain size
         lines.withIndex().filter { it.value.isNotBlank() && !it.value.trimStart().startsWith(';') }.windowed(windowSize, partialWindows = false)
 
-private fun optimizeCmpSequence(linesByFour: List<List<IndexedValue<String>>>): List<Modification> {
-    // when statement (on bytes) generates a sequence of:
-    //	 lda  $ce01,x
-    //	 cmp  #$20
-    //	 beq  check_prog8_s72choice_32
-    //	 lda  $ce01,x
-    //	 cmp  #$21
-    //	 beq  check_prog8_s73choice_33
-    // the repeated lda can be removed
-    val mods = mutableListOf<Modification>()
-    for(lines in linesByFour) {
-        if(lines[0].value.trim()=="lda  P8ESTACK_LO+1,x" &&
-                lines[1].value.trim().startsWith("cmp ") &&
-                lines[2].value.trim().startsWith("beq ") &&
-                lines[3].value.trim()=="lda  P8ESTACK_LO+1,x") {
-            mods.add(Modification(lines[3].index, true, null)) // remove the second lda
-        }
-    }
-    return mods
-}
-
-private fun optimizeUselessStackByteWrites(linesByFour: List<List<IndexedValue<String>>>): List<Modification> {
-    // sta on stack, dex, inx, lda from stack -> eliminate this useless stack byte write
-    // this is a lot harder for word values because the instruction sequence varies.
-    val mods = mutableListOf<Modification>()
-    for(lines in linesByFour) {
-        if(lines[0].value.trim()=="sta  P8ESTACK_LO,x" &&
-                lines[1].value.trim()=="dex" &&
-                lines[2].value.trim()=="inx" &&
-                lines[3].value.trim()=="lda  P8ESTACK_LO,x") {
-            mods.add(Modification(lines[1].index, true, null))
-            mods.add(Modification(lines[2].index, true, null))
-            mods.add(Modification(lines[3].index, true, null))
-        }
-    }
-    return mods
-}
-
 private fun optimizeSameAssignments(
     linesByFourteen: List<List<IndexedValue<String>>>,
     machine: IMachineDefinition,
@@ -517,3 +479,63 @@ private fun optimizeJsrRtsAndOtherCombinations(linesByFour: List<List<IndexedVal
     }
     return mods
 }
+
+private fun optimizeUselessPushPopStack(linesByFour: List<List<IndexedValue<String>>>): List<Modification> {
+    val mods = mutableListOf<Modification>()
+
+    fun optimize(register: Char, lines: List<IndexedValue<String>>) {
+        if(lines[0].value.trimStart().startsWith("ph$register")) {
+            if(lines[2].value.trimStart().startsWith("pl$register")) {
+                val second = lines[1].value.trimStart().take(6).lowercase()
+                if(register!in second
+                    && !second.startsWith("jsr")
+                    && !second.startsWith("pl")
+                    && !second.startsWith("ph")) {
+                    mods.add(Modification(lines[0].index, true, null))
+                    mods.add(Modification(lines[2].index, true, null))
+                }
+            }
+            else if (lines[3].value.trimStart().startsWith("pl$register")) {
+                val second = lines[1].value.trimStart().take(6).lowercase()
+                val third = lines[2].value.trimStart().take(6).lowercase()
+                if(register !in second && register !in third
+                    && !second.startsWith("jsr") && !third.startsWith("jsr")
+                    && !second.startsWith("pl") && !third.startsWith("pl")
+                    && !second.startsWith("ph") && !third.startsWith("ph")) {
+                    mods.add(Modification(lines[0].index, true, null))
+                    mods.add(Modification(lines[3].index, true, null))
+                }
+            }
+        }
+    }
+
+    for (lines in linesByFour) {
+        optimize('a', lines)
+        optimize('x', lines)
+        optimize('y', lines)
+    }
+    return mods
+}
+
+private fun optimizeUnneededTempvarInAdd(linesByFour: List<List<IndexedValue<String>>>): List<Modification> {
+    // sequence:  sta  P8ZP_SCRATCH_XX  / lda  something / clc / adc  P8ZP_SCRATCH_XX
+    // this can be performed without the scratch variable:  clc  /  adc  something
+    val mods = mutableListOf<Modification>()
+
+    for(lines in linesByFour) {
+        val first = lines[0].value.trimStart()
+        val second = lines[1].value.trimStart()
+        val third = lines[2].value.trimStart()
+        val fourth = lines[3].value.trimStart()
+        if(first.startsWith("sta  P8ZP_SCRATCH_") && second.startsWith("lda") && third.startsWith("clc") && fourth.startsWith("adc  P8ZP_SCRATCH_") ) {
+            if(fourth.substring(4)==first.substring(4)) {
+                mods.add(Modification(lines[0].index, false, "    clc"))
+                mods.add(Modification(lines[1].index, false, "    adc  ${second.substring(3).trimStart()}"))
+                mods.add(Modification(lines[2].index, true, null))
+                mods.add(Modification(lines[3].index, true, null))
+            }
+        }
+    }
+
+    return mods
+}

codeGenCpu6502/src/prog8/codegen/cpu6502/AssemblyProgram.kt

@@ -1,14 +1,8 @@
 package prog8.codegen.cpu6502
 
-import com.github.michaelbull.result.Ok
-import com.github.michaelbull.result.Result
-import com.github.michaelbull.result.mapError
 import prog8.code.core.*
 import prog8.code.target.C64Target
-import java.io.File
 import java.nio.file.Path
-import kotlin.io.path.Path
-import kotlin.io.path.isRegularFile
 
 
 internal class AssemblyProgram(
@@ -28,15 +22,20 @@ internal class AssemblyProgram(
         val assemblerCommand: List<String>
 
         when (compTarget.name) {
-            in setOf("c64", "c128", "cx16") -> {
+            in setOf("c64", "c128", "cx16", "pet32") -> {
                 // CBM machines .prg generation.
 
                 // add "-Wlong-branch"  to see warnings about conversion of branch instructions to jumps (default = do this silently)
                 val command = mutableListOf("64tass", "--ascii", "--case-sensitive", "--long-branch",
-                    "-Wall", "-Wno-strict-bool", "-Wno-shadow", // "-Werror",
+                    "-Wall", // "-Wno-strict-bool", "-Werror",
                     "--dump-labels", "--vice-labels", "--labels=$viceMonListFile", "--no-monitor"
                 )
 
+                if(options.warnSymbolShadowing)
+                    command.add("-Wshadow")
+                else
+                    command.add("-Wno-shadow")
+
                 if(options.asmQuiet)
                     command.add("--quiet")
 
@@ -65,10 +64,15 @@ internal class AssemblyProgram(
 
                 // TODO are these options okay for atari?
                 val command = mutableListOf("64tass", "--ascii", "--case-sensitive", "--long-branch",
-                    "-Wall", "-Wno-strict-bool", "-Wno-shadow", // "-Werror",
+                    "-Wall", // "-Werror", "-Wno-strict-bool"
                     "--no-monitor"
                 )
 
+                if(options.warnSymbolShadowing)
+                    command.add("-Wshadow")
+                else
+                    command.add("-Wno-shadow")
+
                 if(options.asmQuiet)
                     command.add("--quiet")
 
@@ -130,18 +134,3 @@ internal class AssemblyProgram(
         viceMonListFile.toFile().appendText(breakpoints.joinToString("\n") + "\n")
     }
 }
-
-
-internal fun loadAsmIncludeFile(filename: String, source: SourceCode): Result<String, NoSuchFileException> {
-    return if (filename.startsWith(SourceCode.libraryFilePrefix)) {
-        return com.github.michaelbull.result.runCatching {
-            SourceCode.Resource("/prog8lib/${filename.substring(SourceCode.libraryFilePrefix.length)}").text
-        }.mapError { NoSuchFileException(File(filename)) }
-    } else {
-        val sib = Path(source.origin).resolveSibling(filename)
-        if (sib.isRegularFile())
-            Ok(SourceCode.File(sib).text)
-        else
-            Ok(SourceCode.File(Path(filename)).text)
-    }
-}

codeGenCpu6502/src/prog8/codegen/cpu6502/BuiltinFunctionsAsmGen.kt

@@ -10,34 +10,31 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                                       private val asmgen: AsmGen6502Internal,
                                       private val assignAsmGen: AssignmentAsmGen) {
 
-    internal fun translateFunctioncallExpression(fcall: PtBuiltinFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?): DataType? {
-        return translateFunctioncall(fcall, discardResult = false, resultToStack = resultToStack, resultRegister = resultRegister)
+    internal fun translateFunctioncallExpression(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?): DataType? {
+        return translateFunctioncall(fcall, discardResult = false, resultRegister = resultRegister)
     }
 
     internal fun translateFunctioncallStatement(fcall: PtBuiltinFunctionCall) {
-        translateFunctioncall(fcall, discardResult = true, resultToStack = false, resultRegister = null)
+        translateFunctioncall(fcall, discardResult = true, resultRegister = null)
     }
 
-    private fun translateFunctioncall(fcall: PtBuiltinFunctionCall, discardResult: Boolean, resultToStack: Boolean, resultRegister: RegisterOrPair?): DataType? {
+    private fun translateFunctioncall(fcall: PtBuiltinFunctionCall, discardResult: Boolean, resultRegister: RegisterOrPair?): DataType? {
         if (discardResult && fcall.hasNoSideEffects)
             return null  // can just ignore the whole function call altogether
 
-        if(discardResult && resultToStack)
-            throw AssemblyError("cannot both discard the result AND put it onto stack")
-
         val sscope = fcall.definingISub()
 
         when (fcall.name) {
-            "msb" -> funcMsb(fcall, resultToStack, resultRegister)
-            "lsb" -> funcLsb(fcall, resultToStack, resultRegister)
-            "mkword" -> funcMkword(fcall, resultToStack, resultRegister)
-            "clamp__byte", "clamp__ubyte", "clamp__word", "clamp__uword" -> funcClamp(fcall, resultToStack, resultRegister)
-            "min__byte", "min__ubyte", "min__word", "min__uword" -> funcMin(fcall, resultToStack, resultRegister)
-            "max__byte", "max__ubyte", "max__word", "max__uword" -> funcMax(fcall, resultToStack, resultRegister)
-            "abs__byte", "abs__word", "abs__float" -> funcAbs(fcall, resultToStack, resultRegister, sscope)
-            "any", "all" -> funcAnyAll(fcall, resultToStack, resultRegister, sscope)
-            "sgn" -> funcSgn(fcall, resultToStack, resultRegister, sscope)
-            "sqrt__ubyte", "sqrt__uword", "sqrt__float" -> funcSqrt(fcall, resultToStack, resultRegister, sscope)
+            "msb" -> funcMsb(fcall, resultRegister)
+            "lsb" -> funcLsb(fcall, resultRegister)
+            "mkword" -> funcMkword(fcall, resultRegister)
+            "clamp__byte", "clamp__ubyte", "clamp__word", "clamp__uword" -> funcClamp(fcall, resultRegister)
+            "min__byte", "min__ubyte", "min__word", "min__uword" -> funcMin(fcall, resultRegister)
+            "max__byte", "max__ubyte", "max__word", "max__uword" -> funcMax(fcall, resultRegister)
+            "abs__byte", "abs__word", "abs__float" -> funcAbs(fcall, resultRegister, sscope)
+            "any", "all" -> funcAnyAll(fcall, resultRegister, sscope)
+            "sgn" -> funcSgn(fcall, resultRegister, sscope)
+            "sqrt__ubyte", "sqrt__uword", "sqrt__float" -> funcSqrt(fcall, resultRegister, sscope)
             "divmod__ubyte" -> funcDivmod(fcall)
             "divmod__uword" -> funcDivmodW(fcall)
             "rol" -> funcRol(fcall)
@@ -46,8 +43,8 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
             "ror2" -> funcRor2(fcall)
             "sort" -> funcSort(fcall)
             "reverse" -> funcReverse(fcall)
-            "memory" -> funcMemory(fcall, discardResult, resultToStack, resultRegister)
-            "peekw" -> funcPeekW(fcall, resultToStack, resultRegister)
+            "memory" -> funcMemory(fcall, discardResult, resultRegister)
+            "peekw" -> funcPeekW(fcall, resultRegister)
             "peek" -> throw AssemblyError("peek() should have been replaced by @()")
             "pokew" -> funcPokeW(fcall)
             "pokemon" -> { /* meme function */ }
@@ -71,18 +68,35 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                 asmgen.popCpuStack(DataType.UWORD, target, fcall.definingISub())
             }
             "rsave" -> funcRsave()
-            "rsavex" -> funcRsaveX()
             "rrestore" -> funcRrestore()
-            "rrestorex" -> funcRrestoreX()
             "cmp" -> funcCmp(fcall)
             "callfar" -> funcCallFar(fcall)
-            "prog8_lib_stringcompare" -> funcStringCompare(fcall, resultToStack)
+            "prog8_lib_stringcompare" -> funcStringCompare(fcall)
+            "prog8_lib_square_byte" -> funcSquare(fcall, DataType.UBYTE)
+            "prog8_lib_square_word" -> funcSquare(fcall, DataType.UWORD)
             else -> throw AssemblyError("missing asmgen for builtin func ${fcall.name}")
         }
 
         return BuiltinFunctions.getValue(fcall.name).returnType
     }
 
+    private fun funcSquare(fcall: PtBuiltinFunctionCall, resultType: DataType) {
+        // square of word value is faster with dedicated routine, square of byte just use the regular multiplication routine.
+        when (resultType) {
+            DataType.UBYTE -> {
+                asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A)
+                asmgen.out("  tay  |  jsr  math.multiply_bytes")
+            }
+            DataType.UWORD -> {
+                asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.AY)
+                asmgen.out("  jsr  math.square")
+            }
+            else -> {
+                throw AssemblyError("optimized square only for integer types")
+            }
+        }
+    }
+
     private fun funcDivmod(fcall: PtBuiltinFunctionCall) {
         assignAsmGen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A, false)
         asmgen.saveRegisterStack(CpuRegister.A, false)
@@ -115,11 +129,9 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
             sty  $remainderVar+1""")
     }
 
-    private fun funcStringCompare(fcall: PtBuiltinFunctionCall, resultToStack: Boolean) {
+    private fun funcStringCompare(fcall: PtBuiltinFunctionCall) {
         asmgen.assignWordOperandsToAYAndVar(fcall.args[0], fcall.args[1], "P8ZP_SCRATCH_W2")
         asmgen.out("  jsr  prog8_lib.strcmp_mem")
-        if(resultToStack)
-            asmgen.out("  sta  P8ESTACK_LO,x |  dex")
     }
 
     private fun funcRsave() {
@@ -142,13 +154,6 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                 lda  P8ZP_SCRATCH_REG""")
     }
 
-    private fun funcRsaveX() {
-        if (asmgen.isTargetCpu(CpuType.CPU65c02))
-            asmgen.out("  phx")
-        else
-            asmgen.out("  txa |  pha")
-    }
-
     private fun funcRrestore() {
         if (asmgen.isTargetCpu(CpuType.CPU65c02))
             asmgen.out("""
@@ -166,13 +171,6 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                 plp""")
     }
 
-    private fun funcRrestoreX() {
-        if (asmgen.isTargetCpu(CpuType.CPU65c02))
-            asmgen.out("  plx")
-        else
-            asmgen.out("  sta  P8ZP_SCRATCH_B1 |  pla |  tax |  lda  P8ZP_SCRATCH_B1")
-    }
-
     private fun funcCallFar(fcall: PtBuiltinFunctionCall) {
         if(asmgen.options.compTarget.name != "cx16")
             throw AssemblyError("callfar only works on cx16 target at this time")
@@ -253,7 +251,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         }
     }
 
-    private fun funcMemory(fcall: PtBuiltinFunctionCall, discardResult: Boolean, resultToStack: Boolean, resultRegister: RegisterOrPair?) {
+    private fun funcMemory(fcall: PtBuiltinFunctionCall, discardResult: Boolean, resultRegister: RegisterOrPair?) {
         if(discardResult)
             throw AssemblyError("should not discard result of memory allocation at $fcall")
         val name = (fcall.args[0] as PtString).value
@@ -263,41 +261,25 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         addressOf.add(slabname)
         addressOf.parent = fcall
         val src = AsmAssignSource(SourceStorageKind.EXPRESSION, program, asmgen, DataType.UWORD, expression = addressOf)
-        val target =
-            if(resultToStack)
-                AsmAssignTarget(TargetStorageKind.STACK,  asmgen, DataType.UWORD, null, fcall.position)
-            else
-                AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, false, fcall.position, null, asmgen)
+        val target = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, false, fcall.position, null, asmgen)
         val assign = AsmAssignment(src, target, program.memsizer, fcall.position)
         asmgen.translateNormalAssignment(assign, fcall.definingISub())
     }
 
-    private fun funcSqrt(fcall: PtBuiltinFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?, scope: IPtSubroutine?) {
+    private fun funcSqrt(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?, scope: IPtSubroutine?) {
         translateArguments(fcall, scope)
         when(fcall.args[0].type) {
             DataType.UBYTE -> {
-                if(resultToStack)
-                    asmgen.out("  ldy  #0 |  jsr  prog8_lib.func_sqrt16_stack")
-                else {
-                    asmgen.out("  ldy  #0 |  jsr  prog8_lib.func_sqrt16_into_A")
-                    assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, false)
-                }
+                asmgen.out("  ldy  #0 |  jsr  prog8_lib.func_sqrt16_into_A")
+                assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, false)
             }
             DataType.UWORD -> {
-                if(resultToStack)
-                    asmgen.out("  jsr  prog8_lib.func_sqrt16_stack")
-                else {
-                    asmgen.out("  jsr  prog8_lib.func_sqrt16_into_A")
-                    assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, false)
-                }
+                asmgen.out("  jsr  prog8_lib.func_sqrt16_into_A")
+                assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, false)
             }
             DataType.FLOAT -> {
-                if(resultToStack)
-                    throw AssemblyError("no support for sqrt float onto stack")
-                else {
-                    asmgen.out("  jsr  floats.func_sqrt_into_FAC1")
-                    assignAsmGen.assignFAC1float(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.FAC1, true, fcall.position, scope, asmgen))
-                }
+                asmgen.out("  jsr  floats.func_sqrt_into_FAC1")
+                assignAsmGen.assignFAC1float(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.FAC1, true, fcall.position, scope, asmgen))
             }
             else -> throw AssemblyError("weird dt")
         }
@@ -466,16 +448,13 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                         } else {
                             val ptrAndIndex = asmgen.pointerViaIndexRegisterPossible(what.address)
                             if(ptrAndIndex!=null) {
-                                asmgen.saveRegisterLocal(CpuRegister.X, fcall.definingISub()!!)
-                                asmgen.assignExpressionToRegister(ptrAndIndex.second, RegisterOrPair.X)
-                                asmgen.saveRegisterLocal(CpuRegister.X, fcall.definingISub()!!)
+                                asmgen.assignExpressionToRegister(ptrAndIndex.second, RegisterOrPair.A)
+                                asmgen.saveRegisterStack(CpuRegister.A, true)
                                 asmgen.assignExpressionToRegister(ptrAndIndex.first, RegisterOrPair.AY)
-                                asmgen.restoreRegisterLocal(CpuRegister.X)
+                                asmgen.out("  sta  (+) + 1 |  sty  (+) + 2")
+                                asmgen.restoreRegisterStack(CpuRegister.X, false)
                                 asmgen.out("""
-                                    sta  (+) + 1
-                                    sty  (+) + 2
 +                                   ror  ${'$'}ffff,x           ; modified""")
-                                asmgen.restoreRegisterLocal(CpuRegister.X)
                             } else {
                                 asmgen.assignExpressionToRegister(what.address, RegisterOrPair.AY)
                                 asmgen.out("""
@@ -571,16 +550,13 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                         } else {
                             val ptrAndIndex = asmgen.pointerViaIndexRegisterPossible(what.address)
                             if(ptrAndIndex!=null) {
-                                asmgen.saveRegisterLocal(CpuRegister.X, fcall.definingISub()!!)
-                                asmgen.assignExpressionToRegister(ptrAndIndex.second, RegisterOrPair.X)
-                                asmgen.saveRegisterLocal(CpuRegister.X, fcall.definingISub()!!)
+                                asmgen.assignExpressionToRegister(ptrAndIndex.second, RegisterOrPair.A)
+                                asmgen.saveRegisterStack(CpuRegister.A, true)
                                 asmgen.assignExpressionToRegister(ptrAndIndex.first, RegisterOrPair.AY)
-                                asmgen.restoreRegisterLocal(CpuRegister.X)
+                                asmgen.out("  sta  (+) + 1 |  sty  (+) + 2")
+                                asmgen.restoreRegisterStack(CpuRegister.X, false)
                                 asmgen.out("""
-                                    sta  (+) + 1
-                                    sty  (+) + 2
 +                                   rol  ${'$'}ffff,x           ; modified""")
-                                asmgen.restoreRegisterLocal(CpuRegister.X)
                             } else {
                                 asmgen.assignExpressionToRegister(what.address, RegisterOrPair.AY)
                                 asmgen.out("""
@@ -633,85 +609,56 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         asmgen.assignExpressionToVariable(indexer.index, "prog8_lib.${operation}_array_u${dt}._arg_index", DataType.UBYTE)
     }
 
-    private fun funcSgn(fcall: PtBuiltinFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?, scope: IPtSubroutine?) {
+    private fun funcSgn(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?, scope: IPtSubroutine?) {
         translateArguments(fcall, scope)
         val dt = fcall.args.single().type
-        if(resultToStack) {
-            when (dt) {
-                DataType.UBYTE -> asmgen.out("  jsr  prog8_lib.func_sign_ub_stack")
-                DataType.BYTE -> asmgen.out("  jsr  prog8_lib.func_sign_b_stack")
-                DataType.UWORD -> asmgen.out("  jsr  prog8_lib.func_sign_uw_stack")
-                DataType.WORD -> asmgen.out("  jsr  prog8_lib.func_sign_w_stack")
-                DataType.FLOAT -> asmgen.out("  jsr  floats.func_sign_f_stack")
-                else -> throw AssemblyError("weird type $dt")
-            }
-        } else {
-            when (dt) {
-                DataType.UBYTE -> asmgen.out("  jsr  prog8_lib.func_sign_ub_into_A")
-                DataType.BYTE -> asmgen.out("  jsr  prog8_lib.func_sign_b_into_A")
-                DataType.UWORD -> asmgen.out("  jsr  prog8_lib.func_sign_uw_into_A")
-                DataType.WORD -> asmgen.out("  jsr  prog8_lib.func_sign_w_into_A")
-                DataType.FLOAT -> asmgen.out("  jsr  floats.func_sign_f_into_A")
-                else -> throw AssemblyError("weird type $dt")
-            }
-            assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, true)
+        when (dt) {
+            DataType.UBYTE -> asmgen.out("  jsr  prog8_lib.func_sign_ub_into_A")
+            DataType.BYTE -> asmgen.out("  jsr  prog8_lib.func_sign_b_into_A")
+            DataType.UWORD -> asmgen.out("  jsr  prog8_lib.func_sign_uw_into_A")
+            DataType.WORD -> asmgen.out("  jsr  prog8_lib.func_sign_w_into_A")
+            DataType.FLOAT -> asmgen.out("  jsr  floats.func_sign_f_into_A")
+            else -> throw AssemblyError("weird type $dt")
         }
+        assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, true)
     }
 
-    private fun funcAnyAll(fcall: PtBuiltinFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?, scope: IPtSubroutine?) {
+    private fun funcAnyAll(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?, scope: IPtSubroutine?) {
         outputAddressAndLenghtOfArray(fcall.args[0])
         val dt = fcall.args.single().type
-        if(resultToStack) {
-            when (dt) {
-                DataType.ARRAY_B, DataType.ARRAY_UB, DataType.STR -> asmgen.out("  jsr  prog8_lib.func_${fcall.name}_b_stack")
-                DataType.ARRAY_UW, DataType.ARRAY_W -> asmgen.out("  jsr  prog8_lib.func_${fcall.name}_w_stack")
-                DataType.ARRAY_F -> asmgen.out("  jsr  floats.func_${fcall.name}_f_stack")
-                in SplitWordArrayTypes -> TODO("split word any/all")
-                else -> throw AssemblyError("weird type $dt")
-            }
-        } else {
-            when (dt) {
-                DataType.ARRAY_B, DataType.ARRAY_UB, DataType.STR -> asmgen.out("  jsr  prog8_lib.func_${fcall.name}_b_into_A |  ldy  #0")
-                DataType.ARRAY_UW, DataType.ARRAY_W -> asmgen.out("  jsr  prog8_lib.func_${fcall.name}_w_into_A |  ldy  #0")
-                DataType.ARRAY_F -> asmgen.out("  jsr  floats.func_${fcall.name}_f_into_A |  ldy  #0")
-                in SplitWordArrayTypes -> TODO("split word any/all")
-                else -> throw AssemblyError("weird type $dt")
-            }
-            assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, dt in SignedDatatypes)
+        when (dt) {
+            DataType.ARRAY_B, DataType.ARRAY_UB, DataType.STR -> asmgen.out("  jsr  prog8_lib.func_${fcall.name}_b_into_A |  ldy  #0")
+            DataType.ARRAY_UW, DataType.ARRAY_W -> asmgen.out("  jsr  prog8_lib.func_${fcall.name}_w_into_A |  ldy  #0")
+            DataType.ARRAY_F -> asmgen.out("  jsr  floats.func_${fcall.name}_f_into_A |  ldy  #0")
+            in SplitWordArrayTypes -> TODO("split word any/all")
+            else -> throw AssemblyError("weird type $dt")
         }
+        assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, dt in SignedDatatypes)
     }
 
-    private fun funcAbs(fcall: PtBuiltinFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?, scope: IPtSubroutine?) {
+    private fun funcAbs(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?, scope: IPtSubroutine?) {
         translateArguments(fcall, scope)
         val dt = fcall.args.single().type
-        if(resultToStack) {
-            when (dt) {
-                DataType.BYTE -> asmgen.out("  jsr  prog8_lib.abs_b_stack")
-                DataType.WORD -> asmgen.out("  jsr  prog8_lib.abs_w_stack")
-                else -> throw AssemblyError("no support for abs onto stack for this dt")
+        when (dt) {
+            DataType.BYTE -> {
+                asmgen.out("  jsr  prog8_lib.abs_b_into_A")
+                assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A,false)
             }
-        } else {
-            when (dt) {
-                DataType.BYTE -> {
-                    asmgen.out("  jsr  prog8_lib.abs_b_into_A")
-                    assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A,false)
-                }
-                DataType.WORD -> {
-                    asmgen.out("  jsr  prog8_lib.abs_w_into_AY")
-                    assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, false, fcall.position, scope, asmgen), RegisterOrPair.AY)
-                }
-                DataType.FLOAT -> {
-                    asmgen.out("  jsr  floats.func_abs_f_into_FAC1")
-                    assignAsmGen.assignFAC1float(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.FAC1, true, fcall.position, scope, asmgen))
-                }
-                DataType.UBYTE -> {
-                    asmgen.assignRegister(RegisterOrPair.A, AsmAssignTarget.fromRegisters(resultRegister?:RegisterOrPair.A, false, fcall.position, scope, asmgen))
-                }
-                DataType.UWORD -> {
-                    asmgen.assignRegister(RegisterOrPair.AY, AsmAssignTarget.fromRegisters(resultRegister?:RegisterOrPair.AY, false, fcall.position, scope, asmgen))
-                }
-                else -> throw AssemblyError("weird type")
+            DataType.WORD -> {
+                asmgen.out("  jsr  prog8_lib.abs_w_into_AY")
+                assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, false, fcall.position, scope, asmgen), RegisterOrPair.AY)
             }
+            DataType.FLOAT -> {
+                asmgen.out("  jsr  floats.func_abs_f_into_FAC1")
+                assignAsmGen.assignFAC1float(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.FAC1, true, fcall.position, scope, asmgen))
+            }
+            DataType.UBYTE -> {
+                asmgen.assignRegister(RegisterOrPair.A, AsmAssignTarget.fromRegisters(resultRegister?:RegisterOrPair.A, false, fcall.position, scope, asmgen))
+            }
+            DataType.UWORD -> {
+                asmgen.assignRegister(RegisterOrPair.AY, AsmAssignTarget.fromRegisters(resultRegister?:RegisterOrPair.AY, false, fcall.position, scope, asmgen))
+            }
+            else -> throw AssemblyError("weird type")
         }
     }
 
@@ -727,7 +674,6 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                 val varname = asmgen.asmVariableName(addrExpr)
                 if(asmgen.isZpVar(addrExpr)) {
                     // pointervar is already in the zero page, no need to copy
-                    asmgen.saveRegisterLocal(CpuRegister.X, fcall.definingISub()!!)
                     asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.AX)
                     if (asmgen.isTargetCpu(CpuType.CPU65c02)) {
                         asmgen.out("""
@@ -743,29 +689,24 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                             iny
                             sta  ($varname),y""")
                     }
-                    asmgen.restoreRegisterLocal(CpuRegister.X)
                     return
                 }
             }
             is PtBinaryExpression -> {
-                require(!asmgen.options.useNewExprCode)
                 val result = asmgen.pointerViaIndexRegisterPossible(addrExpr)
                 val pointer = result?.first as? PtIdentifier
                 if(result!=null && pointer!=null && asmgen.isZpVar(pointer)) {
                     // can do ZP,Y indexing
                     val varname = asmgen.asmVariableName(pointer)
-                    val scope = fcall.definingISub()!!
-                    asmgen.saveRegisterLocal(CpuRegister.X, scope)
                     asmgen.assignExpressionToRegister(result.second, RegisterOrPair.Y)
-                    asmgen.saveRegisterLocal(CpuRegister.Y, scope)
+                    asmgen.saveRegisterStack(CpuRegister.Y, false)
                     asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.AX)
-                    asmgen.restoreRegisterLocal(CpuRegister.Y)
+                    asmgen.restoreRegisterStack(CpuRegister.Y, true)
                     asmgen.out("""
                             sta  ($varname),y
                             txa
                             iny
                             sta  ($varname),y""")
-                    asmgen.restoreRegisterLocal(CpuRegister.X)
                     return
                 }
             }
@@ -777,7 +718,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         asmgen.out("  jsr  prog8_lib.func_pokew")
     }
 
-    private fun funcPeekW(fcall: PtBuiltinFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?) {
+    private fun funcPeekW(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
         fun fallback() {
             asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.AY)
             asmgen.out("  jsr  prog8_lib.func_peekw")
@@ -801,16 +742,15 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                         asmgen.out("""
                             ldy  #0
                             lda  ($varname),y
-                            pha
+                            tax
                             iny
                             lda  ($varname),y
                             tay
-                            pla""")
+                            txa""")
                     }
                 } else fallback()
             }
             is PtBinaryExpression -> {
-                require(!asmgen.options.useNewExprCode)
                 val result = asmgen.pointerViaIndexRegisterPossible(addrExpr)
                 val pointer = result?.first as? PtIdentifier
                 if(result!=null && pointer!=null && asmgen.isZpVar(pointer)) {
@@ -819,33 +759,29 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     asmgen.assignExpressionToRegister(result.second, RegisterOrPair.Y)
                     asmgen.out("""
                         lda  ($varname),y
-                        pha
+                        tax
                         iny
                         lda  ($varname),y
                         tay
-                        pla""")
+                        txa""")
                 } else fallback()
             }
             else -> fallback()
         }
 
-        if(resultToStack){
-            asmgen.out("  sta  P8ESTACK_LO,x |  tya |  sta  P8ESTACK_HI,x |  dex")
-        } else {
-            when(resultRegister ?: RegisterOrPair.AY) {
-                RegisterOrPair.AY -> {}
-                RegisterOrPair.AX -> asmgen.out("  sty  P8ZP_SCRATCH_REG |  ldx  P8ZP_SCRATCH_REG")
-                RegisterOrPair.XY -> asmgen.out("  tax")
-                in Cx16VirtualRegisters -> asmgen.out(
-                    "  sta  cx16.${
-                        resultRegister.toString().lowercase()
-                    } |  sty  cx16.${resultRegister.toString().lowercase()}+1")
-                else -> throw AssemblyError("invalid reg")
-            }
+        when(resultRegister ?: RegisterOrPair.AY) {
+            RegisterOrPair.AY -> {}
+            RegisterOrPair.AX -> asmgen.out("  sty  P8ZP_SCRATCH_REG |  ldx  P8ZP_SCRATCH_REG")
+            RegisterOrPair.XY -> asmgen.out("  tax")
+            in Cx16VirtualRegisters -> asmgen.out(
+                "  sta  cx16.${
+                    resultRegister.toString().lowercase()
+                } |  sty  cx16.${resultRegister.toString().lowercase()}+1")
+            else -> throw AssemblyError("invalid reg")
         }
     }
 
-    private fun funcClamp(fcall: PtBuiltinFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?) {
+    private fun funcClamp(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
         val signed = fcall.type in SignedDatatypes
         when(fcall.type) {
             in ByteDatatypes -> {
@@ -853,30 +789,22 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                 assignAsmGen.assignExpressionToVariable(fcall.args[2], "P8ZP_SCRATCH_W1+1", fcall.args[2].type)  // maximum
                 assignAsmGen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A, signed)    // value
                 asmgen.out("  jsr  prog8_lib.func_clamp_${fcall.type.toString().lowercase()}")
-                if(resultToStack) {
-                    asmgen.out("  sta  P8ESTACK_LO,x |  dex")
-                } else {
-                    val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, signed, fcall.position, fcall.definingISub(), asmgen)
-                    assignAsmGen.assignRegisterByte(targetReg, CpuRegister.A, signed)
-                }
+                val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, signed, fcall.position, fcall.definingISub(), asmgen)
+                assignAsmGen.assignRegisterByte(targetReg, CpuRegister.A, signed)
             }
             in WordDatatypes -> {
                 assignAsmGen.assignExpressionToVariable(fcall.args[1], "P8ZP_SCRATCH_W1", fcall.args[1].type)  // minimum
                 assignAsmGen.assignExpressionToVariable(fcall.args[2], "P8ZP_SCRATCH_W2", fcall.args[2].type)  // maximum
                 assignAsmGen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.AY, signed)    // value
                 asmgen.out("  jsr  prog8_lib.func_clamp_${fcall.type.toString().lowercase()}")
-                if(resultToStack) {
-                    asmgen.out("  sta  P8ESTACK_LO,x |  sty  P8ESTACK_HI,x |  dex")
-                } else {
-                    val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, signed, fcall.position, fcall.definingISub(), asmgen)
-                    assignAsmGen.assignRegisterpairWord(targetReg, RegisterOrPair.AY)
-                }
+                val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, signed, fcall.position, fcall.definingISub(), asmgen)
+                assignAsmGen.assignRegisterpairWord(targetReg, RegisterOrPair.AY)
             }
             else -> throw AssemblyError("invalid dt")
         }
     }
 
-    private fun funcMin(fcall: PtBuiltinFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?) {
+    private fun funcMin(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
         val signed = fcall.type in SignedDatatypes
         if(fcall.type in ByteDatatypes) {
             asmgen.assignExpressionToVariable(fcall.args[1], "P8ZP_SCRATCH_B1", fcall.type)     // right
@@ -886,12 +814,8 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
             asmgen.out("""
                 lda  P8ZP_SCRATCH_B1
 +""")
-            if(resultToStack) {
-                asmgen.out("  sta  P8ESTACK_LO,x |  dex")
-            } else {
-                val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, signed, fcall.position, fcall.definingISub(), asmgen)
-                asmgen.assignRegister(RegisterOrPair.A, targetReg)
-            }
+            val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, signed, fcall.position, fcall.definingISub(), asmgen)
+            asmgen.assignRegister(RegisterOrPair.A, targetReg)
         } else if(fcall.type in WordDatatypes) {
             asmgen.assignExpressionToVariable(fcall.args[0], "P8ZP_SCRATCH_W1", fcall.type)     // left
             asmgen.assignExpressionToVariable(fcall.args[1], "P8ZP_SCRATCH_W2", fcall.type)     // right
@@ -927,18 +851,14 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     ldy  P8ZP_SCRATCH_W1+1
 +""")
             }
-            if(resultToStack) {
-                asmgen.out("  sta  P8ESTACK_LO,x |  sty  P8ESTACK_HI,x |  dex")
-            } else {
-                val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, signed, fcall.position, fcall.definingISub(), asmgen)
-                asmgen.assignRegister(RegisterOrPair.AY, targetReg)
-            }
+            val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, signed, fcall.position, fcall.definingISub(), asmgen)
+            asmgen.assignRegister(RegisterOrPair.AY, targetReg)
         } else {
             throw AssemblyError("min float not supported")
         }
     }
 
-    private fun funcMax(fcall: PtBuiltinFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?) {
+    private fun funcMax(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
         val signed = fcall.type in SignedDatatypes
         if(fcall.type in ByteDatatypes) {
             asmgen.assignExpressionToVariable(fcall.args[0], "P8ZP_SCRATCH_B1", fcall.type)     // left
@@ -948,12 +868,8 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
             asmgen.out("""
                 lda  P8ZP_SCRATCH_B1
 +""")
-            if(resultToStack) {
-                asmgen.out("  sta  P8ESTACK_LO,x |  dex")
-            } else {
-                val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, signed, fcall.position, fcall.definingISub(), asmgen)
-                asmgen.assignRegister(RegisterOrPair.A, targetReg)
-            }
+            val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, signed, fcall.position, fcall.definingISub(), asmgen)
+            asmgen.assignRegister(RegisterOrPair.A, targetReg)
         } else if(fcall.type in WordDatatypes) {
             asmgen.assignExpressionToVariable(fcall.args[0], "P8ZP_SCRATCH_W1", fcall.type)     // left
             asmgen.assignExpressionToVariable(fcall.args[1], "P8ZP_SCRATCH_W2", fcall.type)     // right
@@ -989,72 +905,61 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     ldy  P8ZP_SCRATCH_W2+1
 +""")
             }
-            if(resultToStack) {
-                asmgen.out("  sta  P8ESTACK_LO,x |  sty  P8ESTACK_HI,x |  dex")
-            } else {
-                val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, signed, fcall.position, fcall.definingISub(), asmgen)
-                asmgen.assignRegister(RegisterOrPair.AY, targetReg)
-            }
+            val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, signed, fcall.position, fcall.definingISub(), asmgen)
+            asmgen.assignRegister(RegisterOrPair.AY, targetReg)
         } else {
             throw AssemblyError("max float not supported")
         }
     }
 
-    private fun funcMkword(fcall: PtBuiltinFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?) {
-        if(resultToStack) {
-            asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A)      // msb
-            asmgen.out("  pha")
-            asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)      // lsb
-            asmgen.out("  sta  P8ESTACK_LO,x |  pla |  sta  P8ESTACK_HI,x |  dex")
-        } else {
-            val reg = resultRegister ?: RegisterOrPair.AY
-            var needAsave = asmgen.needAsaveForExpr(fcall.args[0])
-            if(!needAsave) {
-                val mr0 = fcall.args[0] as? PtMemoryByte
-                val mr1 = fcall.args[1] as? PtMemoryByte
-                if (mr0 != null)
-                    needAsave =  mr0.address !is PtNumber
-                if (mr1 != null)
-                    needAsave = needAsave or (mr1.address !is PtNumber)
+    private fun funcMkword(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
+        val reg = resultRegister ?: RegisterOrPair.AY
+        var needAsave = asmgen.needAsaveForExpr(fcall.args[0])
+        if(!needAsave) {
+            val mr0 = fcall.args[0] as? PtMemoryByte
+            val mr1 = fcall.args[1] as? PtMemoryByte
+            if (mr0 != null)
+                needAsave =  mr0.address !is PtNumber
+            if (mr1 != null)
+                needAsave = needAsave or (mr1.address !is PtNumber)
+        }
+        when(reg) {
+            RegisterOrPair.AX -> {
+                asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)      // lsb
+                if(needAsave)
+                    asmgen.out("  pha")
+                asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.X)      // msb
+                if(needAsave)
+                    asmgen.out("  pla")
             }
-            when(reg) {
-                RegisterOrPair.AX -> {
-                    asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)      // lsb
-                    if(needAsave)
-                        asmgen.out("  pha")
-                    asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.X)      // msb
-                    if(needAsave)
-                        asmgen.out("  pla")
-                }
-                RegisterOrPair.AY -> {
-                    asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)      // lsb
-                    if(needAsave)
-                        asmgen.out("  pha")
-                    asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.Y)      // msb
-                    if(needAsave)
-                        asmgen.out("  pla")
-                }
-                RegisterOrPair.XY -> {
-                    asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)      // lsb
-                    if(needAsave)
-                        asmgen.out("  pha")
-                    asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.Y)      // msb
-                    if(needAsave)
-                        asmgen.out("  pla")
-                    asmgen.out("  tax")
-                }
-                in Cx16VirtualRegisters -> {
-                    asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)      // lsb
-                    asmgen.out("  sta  cx16.${reg.toString().lowercase()}")
-                    asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A)      // msb
-                    asmgen.out("  sta  cx16.${reg.toString().lowercase()}+1")
-                }
-                else -> throw AssemblyError("invalid mkword target reg")
+            RegisterOrPair.AY -> {
+                asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)      // lsb
+                if(needAsave)
+                    asmgen.out("  pha")
+                asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.Y)      // msb
+                if(needAsave)
+                    asmgen.out("  pla")
             }
+            RegisterOrPair.XY -> {
+                asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)      // lsb
+                if(needAsave)
+                    asmgen.out("  pha")
+                asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.Y)      // msb
+                if(needAsave)
+                    asmgen.out("  pla")
+                asmgen.out("  tax")
+            }
+            in Cx16VirtualRegisters -> {
+                asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)      // lsb
+                asmgen.out("  sta  cx16.${reg.toString().lowercase()}")
+                asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A)      // msb
+                asmgen.out("  sta  cx16.${reg.toString().lowercase()}+1")
+            }
+            else -> throw AssemblyError("invalid mkword target reg")
         }
     }
 
-    private fun funcMsb(fcall: PtBuiltinFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?) {
+    private fun funcMsb(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
         val arg = fcall.args.single()
         if (arg.type !in WordDatatypes)
             throw AssemblyError("msb required word argument")
@@ -1062,53 +967,44 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
             throw AssemblyError("msb(const) should have been const-folded away")
         if (arg is PtIdentifier) {
             val sourceName = asmgen.asmVariableName(arg)
-            if(resultToStack) {
-                asmgen.out("  lda  $sourceName+1 |  sta  P8ESTACK_LO,x |  dex")
-            } else {
-                when(resultRegister) {
-                    null, RegisterOrPair.A -> asmgen.out("  lda  $sourceName+1")
-                    RegisterOrPair.X -> asmgen.out("  ldx  $sourceName+1")
-                    RegisterOrPair.Y -> asmgen.out("  ldy  $sourceName+1")
-                    RegisterOrPair.AX -> asmgen.out("  lda  $sourceName+1 |  ldx  #0")
-                    RegisterOrPair.AY -> asmgen.out("  lda  $sourceName+1 |  ldy  #0")
-                    RegisterOrPair.XY -> asmgen.out("  ldx  $sourceName+1 |  ldy  #0")
-                    in Cx16VirtualRegisters -> {
-                        val regname = resultRegister.name.lowercase()
-                        if(asmgen.isTargetCpu(CpuType.CPU65c02))
-                            asmgen.out("  lda  $sourceName+1 |  sta  cx16.$regname |  stz  cx16.$regname+1")
-                        else
-                            asmgen.out("  lda  $sourceName+1 |  sta  cx16.$regname |  lda  #0 |  sta  cx16.$regname+1")
-                    }
-                    else -> throw AssemblyError("invalid reg")
+            when(resultRegister) {
+                null, RegisterOrPair.A -> asmgen.out("  lda  $sourceName+1")
+                RegisterOrPair.X -> asmgen.out("  ldx  $sourceName+1")
+                RegisterOrPair.Y -> asmgen.out("  ldy  $sourceName+1")
+                RegisterOrPair.AX -> asmgen.out("  lda  $sourceName+1 |  ldx  #0")
+                RegisterOrPair.AY -> asmgen.out("  lda  $sourceName+1 |  ldy  #0")
+                RegisterOrPair.XY -> asmgen.out("  ldx  $sourceName+1 |  ldy  #0")
+                in Cx16VirtualRegisters -> {
+                    val regname = resultRegister.name.lowercase()
+                    if(asmgen.isTargetCpu(CpuType.CPU65c02))
+                        asmgen.out("  lda  $sourceName+1 |  sta  cx16.$regname |  stz  cx16.$regname+1")
+                    else
+                        asmgen.out("  lda  $sourceName+1 |  sta  cx16.$regname |  lda  #0 |  sta  cx16.$regname+1")
                 }
+                else -> throw AssemblyError("invalid reg")
             }
         } else {
-            if(resultToStack) {
-                asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
-                asmgen.out("  tya |  sta  P8ESTACK_LO,x |  dex")
-            } else {
-                when(resultRegister) {
-                    null, RegisterOrPair.A -> {
-                        asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
-                        asmgen.out("  tya")
-                    }
-                    RegisterOrPair.X -> {
-                        asmgen.out("  pha")
-                        asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AX)
-                        asmgen.out("  pla")
-                    }
-                    RegisterOrPair.Y -> {
-                        asmgen.out("  pha")
-                        asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
-                        asmgen.out("  pla")
-                    }
-                    else -> throw AssemblyError("invalid reg")
+            when(resultRegister) {
+                null, RegisterOrPair.A -> {
+                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
+                    asmgen.out("  tya")
                 }
+                RegisterOrPair.X -> {
+                    asmgen.out("  pha")
+                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AX)
+                    asmgen.out("  pla")
+                }
+                RegisterOrPair.Y -> {
+                    asmgen.out("  pha")
+                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
+                    asmgen.out("  pla")
+                }
+                else -> throw AssemblyError("invalid reg")
             }
         }
     }
 
-    private fun funcLsb(fcall: PtBuiltinFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?) {
+    private fun funcLsb(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
         val arg = fcall.args.single()
         if (arg.type !in WordDatatypes)
             throw AssemblyError("lsb required word argument")
@@ -1117,54 +1013,60 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
 
         if (arg is PtIdentifier) {
             val sourceName = asmgen.asmVariableName(arg)
-            if(resultToStack) {
-                asmgen.out("  lda  $sourceName |  sta  P8ESTACK_LO,x |  dex")
-            } else {
-                when(resultRegister) {
-                    null, RegisterOrPair.A -> asmgen.out("  lda  $sourceName")
-                    RegisterOrPair.X -> asmgen.out("  ldx  $sourceName")
-                    RegisterOrPair.Y -> asmgen.out("  ldy  $sourceName")
-                    RegisterOrPair.AX -> asmgen.out("  lda  $sourceName |  ldx  #0")
-                    RegisterOrPair.AY -> asmgen.out("  lda  $sourceName |  ldy  #0")
-                    RegisterOrPair.XY -> asmgen.out("  ldx  $sourceName |  ldy  #0")
-                    in Cx16VirtualRegisters -> {
-                        val regname = resultRegister.name.lowercase()
-                        if(asmgen.isTargetCpu(CpuType.CPU65c02))
-                            asmgen.out("  lda  $sourceName |  sta  cx16.$regname |  stz  cx16.$regname+1")
-                        else
-                            asmgen.out("  lda  $sourceName |  sta  cx16.$regname |  lda  #0 |  sta  cx16.$regname+1")
-                    }
-                    else -> throw AssemblyError("invalid reg")
+            when(resultRegister) {
+                null, RegisterOrPair.A -> asmgen.out("  lda  $sourceName")
+                RegisterOrPair.X -> asmgen.out("  ldx  $sourceName")
+                RegisterOrPair.Y -> asmgen.out("  ldy  $sourceName")
+                RegisterOrPair.AX -> asmgen.out("  lda  $sourceName |  ldx  #0")
+                RegisterOrPair.AY -> asmgen.out("  lda  $sourceName |  ldy  #0")
+                RegisterOrPair.XY -> asmgen.out("  ldx  $sourceName |  ldy  #0")
+                in Cx16VirtualRegisters -> {
+                    val regname = resultRegister.name.lowercase()
+                    if(asmgen.isTargetCpu(CpuType.CPU65c02))
+                        asmgen.out("  lda  $sourceName |  sta  cx16.$regname |  stz  cx16.$regname+1")
+                    else
+                        asmgen.out("  lda  $sourceName |  sta  cx16.$regname |  lda  #0 |  sta  cx16.$regname+1")
                 }
+                else -> throw AssemblyError("invalid reg")
             }
         } else {
-            if(resultToStack) {
-                asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
-                // NOTE: we rely on the fact that the above assignment to AY, assigns the Lsb to A as the last instruction.
-                //       this is required because the compiler assumes the status bits are set according to what A is (lsb)
-                //       and will not generate another cmp when lsb() is directly used inside a comparison expression.
-                asmgen.out("  sta  P8ESTACK_LO,x |  dex")
-            } else {
-                when(resultRegister) {
-                    null, RegisterOrPair.A -> {
-                        asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
-                        // NOTE: we rely on the fact that the above assignment to AY, assigns the Lsb to A as the last instruction.
-                        //       this is required because the compiler assumes the status bits are set according to what A is (lsb)
-                        //       and will not generate another cmp when lsb() is directly used inside a comparison expression.
-                    }
-                    RegisterOrPair.X -> {
-                        asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.XY)
-                        // NOTE: we rely on the fact that the above assignment to XY, assigns the Lsb to X as the last instruction.
-                        //       this is required because the compiler assumes the status bits are set according to what X is (lsb)
-                        //       and will not generate another cmp when lsb() is directly used inside a comparison expression.
-                    }
-                    RegisterOrPair.Y -> {
-                        asmgen.out("  pha")
-                        asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
-                        asmgen.out("  tay |  pla |  cpy  #0")
-                    }
-                    else -> throw AssemblyError("invalid reg")
+            when(resultRegister) {
+                null, RegisterOrPair.A -> {
+                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
+                    // NOTE: we rely on the fact that the above assignment to AY, assigns the Lsb to A as the last instruction.
+                    //       this is required because the compiler assumes the status bits are set according to what A is (lsb)
+                    //       and will not generate another cmp when lsb() is directly used inside a comparison expression.
                 }
+                RegisterOrPair.X -> {
+                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.XY)
+                    // NOTE: we rely on the fact that the above assignment to XY, assigns the Lsb to X as the last instruction.
+                    //       this is required because the compiler assumes the status bits are set according to what X is (lsb)
+                    //       and will not generate another cmp when lsb() is directly used inside a comparison expression.
+                }
+                RegisterOrPair.Y -> {
+                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
+                    asmgen.out("  tay |  cpy  #0")
+                }
+                RegisterOrPair.AY -> {
+                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
+                    asmgen.out("  ldy  #0 |  cmp  #0")
+                }
+                RegisterOrPair.AX -> {
+                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AX)
+                    asmgen.out("  ldx  #0 |  cmp  #0")
+                }
+                RegisterOrPair.XY -> {
+                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.XY)
+                    asmgen.out("  ldy  #0 |  cpx  #0")
+                }
+                in Cx16VirtualRegisters -> {
+                    asmgen.assignExpressionToRegister(fcall.args.single(), resultRegister)
+                    val zero = PtNumber(DataType.UBYTE, 0.0, Position.DUMMY)
+                    zero.parent=fcall
+                    assignAsmGen.assignExpressionToVariable(zero, "cx16.${resultRegister.toString().lowercase()}H", DataType.UBYTE)
+                    asmgen.out("  lda  cx16.r0L")
+                }
+                else -> throw AssemblyError("invalid reg")
             }
         }
     }

codeGenCpu6502/src/prog8/codegen/cpu6502/ExpressionsAsmGen.kt

@@ -1,935 +0,0 @@
-package prog8.codegen.cpu6502
-
-import prog8.code.ast.*
-import prog8.code.core.*
-import kotlin.math.absoluteValue
-
-internal class ExpressionsAsmGen(private val program: PtProgram,
-                                 private val asmgen: AsmGen6502Internal,
-                                 private val allocator: VariableAllocator) {
-
-    @Deprecated("avoid calling this as it generates slow evalstack based code")
-    internal fun translateExpression(expression: PtExpression) {
-        if (this.asmgen.options.slowCodegenWarnings) {
-            asmgen.errors.warn("slow stack evaluation used for expression", expression.position)
-        }
-        translateExpressionInternal(expression)
-    }
-
-
-    // the rest of the methods are all PRIVATE
-
-
-    private fun translateExpressionInternal(expression: PtExpression) {
-
-        when(expression) {
-            is PtPrefix -> translateExpression(expression)
-            is PtBinaryExpression -> translateExpression(expression)
-            is PtArrayIndexer -> translateExpression(expression)
-            is PtTypeCast -> translateExpression(expression)
-            is PtAddressOf -> translateExpression(expression)
-            is PtMemoryByte -> asmgen.translateDirectMemReadExpressionToRegAorStack(expression, true)
-            is PtNumber -> translateExpression(expression)
-            is PtIdentifier -> translateExpression(expression)
-            is PtFunctionCall -> translateFunctionCallResultOntoStack(expression)
-            is PtBuiltinFunctionCall -> asmgen.translateBuiltinFunctionCallExpression(expression, true, null)
-            is PtContainmentCheck -> translateContainmentCheck(expression)
-            is PtArray, is PtString -> throw AssemblyError("string/array literal value assignment should have been replaced by a variable")
-            is PtRange -> throw AssemblyError("range expression should have been changed into array values")
-            is PtMachineRegister -> throw AssemblyError("machine register ast node should not occur in 6502 codegen it is for IR code")
-            else -> TODO("missing expression asmgen for $expression")
-        }
-    }
-
-    private fun translateContainmentCheck(check: PtContainmentCheck) {
-        asmgen.assignExpressionToRegister(check, RegisterOrPair.A)
-        asmgen.out("  sta  P8ESTACK_LO,x |  dex")
-    }
-
-    private fun translateFunctionCallResultOntoStack(call: PtFunctionCall) {
-        // only for use in nested expression evaluation
-
-        val symbol = asmgen.symbolTable.lookup(call.name)
-        val sub = symbol!!.astNode as IPtSubroutine
-        asmgen.saveXbeforeCall(call)
-        asmgen.translateFunctionCall(call)
-        if(sub.regXasResult()) {
-            // store the return value in X somewhere that we can access again below
-            asmgen.out("  stx  P8ZP_SCRATCH_REG")
-        }
-        asmgen.restoreXafterCall(call)
-
-        val returns: List<Pair<RegisterOrStatusflag, DataType>> = sub.returnsWhatWhere()
-        for ((reg, _) in returns) {
-            // result value is in cpu or status registers, put it on the stack instead (as we're evaluating an expression tree)
-            if (reg.registerOrPair != null) {
-                when (reg.registerOrPair!!) {
-                    RegisterOrPair.A -> asmgen.out("  sta  P8ESTACK_LO,x |  dex")
-                    RegisterOrPair.Y -> asmgen.out("  tya |  sta  P8ESTACK_LO,x |  dex")
-                    RegisterOrPair.AY -> asmgen.out("  sta  P8ESTACK_LO,x |  tya |  sta  P8ESTACK_HI,x |  dex")
-                    RegisterOrPair.X -> asmgen.out("  lda  P8ZP_SCRATCH_REG |  sta  P8ESTACK_LO,x |  dex")
-                    RegisterOrPair.AX -> asmgen.out("  sta  P8ESTACK_LO,x |  lda  P8ZP_SCRATCH_REG |  sta  P8ESTACK_HI,x |  dex")
-                    RegisterOrPair.XY -> asmgen.out("  tya |  sta  P8ESTACK_HI,x |  lda  P8ZP_SCRATCH_REG |  sta  P8ESTACK_LO,x |  dex")
-                    RegisterOrPair.FAC1 -> asmgen.out("  jsr  floats.push_fac1")
-                    RegisterOrPair.FAC2 -> asmgen.out("  jsr  floats.push_fac2")
-                    RegisterOrPair.R0,
-                    RegisterOrPair.R1,
-                    RegisterOrPair.R2,
-                    RegisterOrPair.R3,
-                    RegisterOrPair.R4,
-                    RegisterOrPair.R5,
-                    RegisterOrPair.R6,
-                    RegisterOrPair.R7,
-                    RegisterOrPair.R8,
-                    RegisterOrPair.R9,
-                    RegisterOrPair.R10,
-                    RegisterOrPair.R11,
-                    RegisterOrPair.R12,
-                    RegisterOrPair.R13,
-                    RegisterOrPair.R14,
-                    RegisterOrPair.R15 -> {
-                        asmgen.out(
-                            """
-                            lda  cx16.${reg.registerOrPair.toString().lowercase()}
-                            sta  P8ESTACK_LO,x
-                            lda  cx16.${reg.registerOrPair.toString().lowercase()}+1
-                            sta  P8ESTACK_HI,x
-                            dex
-                        """)
-                    }
-                }
-            } else when(reg.statusflag) {
-                Statusflag.Pc -> {
-                    asmgen.out("""
-                        lda  #0
-                        rol  a
-                        sta  P8ESTACK_LO,x
-                        dex""")
-                }
-                Statusflag.Pz -> {
-                    asmgen.out("""
-                        beq  +
-                        lda  #0
-                        beq  ++
-+                           lda  #1
-+                           sta  P8ESTACK_LO,x
-                        dex""")
-                }
-                Statusflag.Pv -> {
-                    asmgen.out("""
-                        bvs  +
-                        lda  #0
-                        beq  ++
-+                           lda  #1
-+                           sta  P8ESTACK_LO,x
-                        dex""")
-                }
-                Statusflag.Pn -> {
-                    asmgen.out("""
-                        bmi  +
-                        lda  #0
-                        beq  ++
-+                           lda  #1
-+                           sta  P8ESTACK_LO,x
-                        dex""")
-                }
-                null -> {}
-            }
-        }
-    }
-
-    private fun translateExpression(typecast: PtTypeCast) {
-        translateExpressionInternal(typecast.value)
-        when(typecast.value.type) {
-            DataType.UBYTE, DataType.BOOL -> {
-                when(typecast.type) {
-                    DataType.UBYTE, DataType.BYTE -> {}
-                    DataType.UWORD, DataType.WORD -> {
-                        if(asmgen.isTargetCpu(CpuType.CPU65c02))
-                            asmgen.out("  stz  P8ESTACK_HI+1,x")
-                        else
-                            asmgen.out("  lda  #0  |  sta  P8ESTACK_HI+1,x")
-                    }
-                    DataType.FLOAT -> asmgen.out(" jsr  floats.stack_ub2float")
-                    in PassByReferenceDatatypes -> throw AssemblyError("cannot cast to a pass-by-reference datatype")
-                    else -> throw AssemblyError("weird type")
-                }
-            }
-            DataType.BYTE -> {
-                when(typecast.type) {
-                    DataType.UBYTE, DataType.BYTE -> {}
-                    DataType.UWORD, DataType.WORD -> asmgen.signExtendStackLsb(DataType.BYTE)
-                    DataType.FLOAT -> asmgen.out(" jsr  floats.stack_b2float")
-                    in PassByReferenceDatatypes -> throw AssemblyError("cannot cast to a pass-by-reference datatype")
-                    else -> throw AssemblyError("weird type")
-                }
-            }
-            DataType.UWORD -> {
-                when(typecast.type) {
-                    DataType.BYTE, DataType.UBYTE -> {}
-                    DataType.WORD, DataType.UWORD -> {}
-                    DataType.FLOAT -> asmgen.out(" jsr  floats.stack_uw2float")
-                    in PassByReferenceDatatypes -> throw AssemblyError("cannot cast to a pass-by-reference datatype")
-                    else -> throw AssemblyError("weird type")
-                }
-            }
-            DataType.WORD -> {
-                when(typecast.type) {
-                    DataType.BYTE, DataType.UBYTE -> {}
-                    DataType.WORD, DataType.UWORD -> {}
-                    DataType.FLOAT -> asmgen.out(" jsr  floats.stack_w2float")
-                    in PassByReferenceDatatypes -> throw AssemblyError("cannot cast to a pass-by-reference datatype")
-                    else -> throw AssemblyError("weird type")
-                }
-            }
-            DataType.FLOAT -> {
-                when(typecast.type) {
-                    DataType.UBYTE -> asmgen.out(" jsr  floats.stack_float2uw")
-                    DataType.BYTE -> asmgen.out(" jsr  floats.stack_float2w")
-                    DataType.UWORD -> asmgen.out(" jsr  floats.stack_float2uw")
-                    DataType.WORD -> asmgen.out(" jsr  floats.stack_float2w")
-                    DataType.FLOAT -> {}
-                    in PassByReferenceDatatypes -> throw AssemblyError("cannot cast to a pass-by-reference datatype")
-                    else -> throw AssemblyError("weird type")
-                }
-            }
-            DataType.STR -> {
-                if (typecast.type != DataType.UWORD && typecast.type == DataType.STR)
-                    throw AssemblyError("cannot typecast a string into another incompatitble type")
-            }
-            in PassByReferenceDatatypes -> throw AssemblyError("cannot cast pass-by-reference value into another type")
-            else -> throw AssemblyError("weird type")
-        }
-    }
-
-    private fun translateExpression(expr: PtAddressOf) {
-        val name = asmgen.asmVariableName(expr.identifier)
-        asmgen.out("  lda  #<$name |  sta  P8ESTACK_LO,x |  lda  #>$name  |  sta  P8ESTACK_HI,x  | dex")
-    }
-
-    private fun translateExpression(expr: PtNumber) {
-        when(expr.type) {
-            DataType.UBYTE, DataType.BYTE -> asmgen.out(" lda  #${expr.number.toHex()}  | sta  P8ESTACK_LO,x  | dex")
-            DataType.UWORD, DataType.WORD -> asmgen.out("""
-                lda  #<${expr.number.toHex()}
-                sta  P8ESTACK_LO,x
-                lda  #>${expr.number.toHex()}
-                sta  P8ESTACK_HI,x
-                dex
-            """)
-            DataType.FLOAT -> {
-                val floatConst = allocator.getFloatAsmConst(expr.number)
-                asmgen.out(" lda  #<$floatConst |  ldy  #>$floatConst |  jsr  floats.push_float")
-            }
-            else -> throw AssemblyError("weird type")
-        }
-    }
-
-    private fun translateExpression(expr: PtIdentifier) {
-        val varname = asmgen.asmVariableName(expr)
-        when(expr.type) {
-            DataType.UBYTE, DataType.BYTE -> {
-                asmgen.out("  lda  $varname  |  sta  P8ESTACK_LO,x  |  dex")
-            }
-            DataType.UWORD, DataType.WORD -> {
-                asmgen.out("  lda  $varname  |  sta  P8ESTACK_LO,x  |  lda  $varname+1 |  sta  P8ESTACK_HI,x |  dex")
-            }
-            DataType.FLOAT -> {
-                asmgen.out(" lda  #<$varname |  ldy  #>$varname|  jsr  floats.push_float")
-            }
-            in SplitWordArrayTypes -> {
-                throw AssemblyError("can't push address of split-word array ${expr.position}")
-            }
-            in IterableDatatypes -> {
-                asmgen.out("  lda  #<$varname  |  sta  P8ESTACK_LO,x  |  lda  #>$varname |  sta  P8ESTACK_HI,x |  dex")
-            }
-            else -> throw AssemblyError("stack push weird variable type $expr")
-        }
-    }
-
-    private fun translateExpression(expr: PtBinaryExpression) {
-        require(!asmgen.options.useNewExprCode)
-
-        // Uses evalstack to evaluate the given expression.  THIS IS SLOW AND SHOULD BE AVOIDED!
-        if(translateSomewhatOptimized(expr.left, expr.operator, expr.right))
-            return
-
-        val leftDt = expr.left.type
-        val rightDt = expr.right.type
-
-        // compare with zero
-        if(expr.operator in ComparisonOperators) {
-            if(leftDt in NumericDatatypes && rightDt in NumericDatatypes) {
-                val rightVal = expr.right.asConstInteger()
-                if(rightVal==0)
-                    return translateComparisonWithZero(expr.left, leftDt, expr.operator)
-            }
-        }
-
-        if(leftDt==DataType.STR && rightDt==DataType.STR && expr.operator in ComparisonOperators)
-            return translateCompareStrings(expr.left, expr.operator, expr.right)
-
-        if((leftDt in ByteDatatypes && rightDt !in ByteDatatypes)
-                || (leftDt in WordDatatypes && rightDt !in WordDatatypes))
-            throw AssemblyError("binary operator ${expr.operator} left/right dt not identical")
-
-        // the general, non-optimized cases
-        // TODO optimize more cases.... (or one day just don't use the evalstack at all anymore)
-        translateExpressionInternal(expr.left)
-        translateExpressionInternal(expr.right)
-        when (leftDt) {
-            in ByteDatatypes -> translateBinaryOperatorBytes(expr.operator, leftDt)
-            in WordDatatypes -> translateBinaryOperatorWords(expr.operator, leftDt)
-            DataType.FLOAT -> translateBinaryOperatorFloats(expr.operator)
-            else -> throw AssemblyError("non-numerical datatype")
-        }
-    }
-
-    private fun translateSomewhatOptimized(left: PtExpression, operator: String, right: PtExpression): Boolean {
-        val leftDt = left.type
-        val rightDt = right.type
-        when(operator) {
-            "+" -> {
-                if(leftDt in IntegerDatatypes && rightDt in IntegerDatatypes) {
-                    val leftVal = left.asConstInteger()
-                    val rightVal = right.asConstInteger()
-                    if (leftVal!=null && leftVal in -4..4) {
-                        translateExpressionInternal(right)
-                        if(rightDt in ByteDatatypes) {
-                            val incdec = if(leftVal<0) "dec" else "inc"
-                            repeat(leftVal.absoluteValue) {
-                                asmgen.out("  $incdec  P8ESTACK_LO+1,x")
-                            }
-                        } else {
-                            // word
-                            if(leftVal<0) {
-                                repeat(leftVal.absoluteValue) {
-                                    asmgen.out("""
-                                        lda  P8ESTACK_LO+1,x
-                                        bne  +
-                                        dec  P8ESTACK_HI+1,x
-+                                       dec  P8ESTACK_LO+1,x""")
-                                }
-                            } else {
-                                repeat(leftVal) {
-                                    asmgen.out("""
-                                        inc  P8ESTACK_LO+1,x
-                                        bne  +
-                                        inc  P8ESTACK_HI+1,x
-+""")
-                                }
-                            }
-                        }
-                        return true
-                    }
-                    else if (rightVal!=null && rightVal in -4..4)
-                    {
-                        translateExpressionInternal(left)
-                        if(leftDt in ByteDatatypes) {
-                            val incdec = if(rightVal<0) "dec" else "inc"
-                            repeat(rightVal.absoluteValue) {
-                                asmgen.out("  $incdec  P8ESTACK_LO+1,x")
-                            }
-                        } else {
-                            // word
-                            if(rightVal<0) {
-                                repeat(rightVal.absoluteValue) {
-                                    asmgen.out("""
-                                        lda  P8ESTACK_LO+1,x
-                                        bne  +
-                                        dec  P8ESTACK_HI+1,x
-+                                       dec  P8ESTACK_LO+1,x""")
-                                }
-                            } else {
-                                repeat(rightVal) {
-                                    asmgen.out("""
-                                        inc  P8ESTACK_LO+1,x
-                                        bne  +
-                                        inc  P8ESTACK_HI+1,x
-+""")
-                                }
-                            }
-                        }
-                        return true
-                    }
-                }
-            }
-            "-" -> {
-                if(leftDt in IntegerDatatypes && rightDt in IntegerDatatypes) {
-                    val rightVal = right.asConstInteger()
-                    if (rightVal!=null && rightVal in -4..4)
-                    {
-                        translateExpressionInternal(left)
-                        if(leftDt in ByteDatatypes) {
-                            val incdec = if(rightVal<0) "inc" else "dec"
-                            repeat(rightVal.absoluteValue) {
-                                asmgen.out("  $incdec  P8ESTACK_LO+1,x")
-                            }
-                        } else {
-                            // word
-                            if(rightVal>0) {
-                                repeat(rightVal.absoluteValue) {
-                                    asmgen.out("""
-                                        lda  P8ESTACK_LO+1,x
-                                        bne  +
-                                        dec  P8ESTACK_HI+1,x
-+                                       dec  P8ESTACK_LO+1,x""")
-                                }
-                            } else {
-                                repeat(rightVal) {
-                                    asmgen.out("""
-                                        inc  P8ESTACK_LO+1,x
-                                        bne  +
-                                        inc  P8ESTACK_HI+1,x
-+""")
-                                }
-                            }
-                        }
-                        return true
-                    }
-                }
-            }
-            ">>" -> {
-                val amount = right.asConstInteger()
-                if(amount!=null) {
-                    translateExpressionInternal(left)
-                    when (leftDt) {
-                        DataType.UBYTE -> {
-                            if (amount <= 2)
-                                repeat(amount) { asmgen.out(" lsr  P8ESTACK_LO+1,x") }
-                            else {
-                                asmgen.out(" lda  P8ESTACK_LO+1,x")
-                                repeat(amount) { asmgen.out(" lsr  a") }
-                                asmgen.out(" sta  P8ESTACK_LO+1,x")
-                            }
-                        }
-                        DataType.BYTE -> {
-                            if (amount <= 2)
-                                repeat(amount) { asmgen.out(" lda  P8ESTACK_LO+1,x |  asl  a |  ror  P8ESTACK_LO+1,x") }
-                            else {
-                                asmgen.out(" lda  P8ESTACK_LO+1,x |  sta  P8ZP_SCRATCH_B1")
-                                repeat(amount) { asmgen.out(" asl  a |  ror  P8ZP_SCRATCH_B1 |  lda  P8ZP_SCRATCH_B1") }
-                                asmgen.out(" sta  P8ESTACK_LO+1,x")
-                            }
-                        }
-                        DataType.UWORD -> {
-                            if(amount>=16) {
-                                if(asmgen.isTargetCpu(CpuType.CPU65c02))
-                                    asmgen.out("  stz  P8ESTACK_LO+1,x |  stz  P8ESTACK_HI+1,x")
-                                else
-                                    asmgen.out("  lda  #0 |  sta  P8ESTACK_LO+1,x |  sta  P8ESTACK_HI+1,x")
-                                return true
-                            }
-                            var amountLeft = amount
-                            while (amountLeft >= 7) {
-                                asmgen.out(" jsr  math.shift_right_uw_7")
-                                amountLeft -= 7
-                            }
-                            if (amountLeft in 0..2)
-                                repeat(amountLeft) { asmgen.out(" lsr  P8ESTACK_HI+1,x |  ror  P8ESTACK_LO+1,x") }
-                            else
-                                asmgen.out(" jsr  math.shift_right_uw_$amountLeft")
-                        }
-                        DataType.WORD -> {
-                            if(amount>=16) {
-                                asmgen.out("""
-                                    lda  P8ESTACK_HI+1,x
-                                    bmi  +
-                                    lda  #0
-                                    sta  P8ESTACK_LO+1,x
-                                    sta  P8ESTACK_HI+1,x
-                                    beq  ++
-+                                   lda  #255
-                                    sta  P8ESTACK_LO+1,x
-                                    sta  P8ESTACK_HI+1,x
-+""")
-                                return true
-                            }
-                            var amountLeft = amount
-                            while (amountLeft >= 7) {
-                                asmgen.out(" jsr  math.shift_right_w_7")
-                                amountLeft -= 7
-                            }
-                            if (amountLeft in 0..2)
-                                repeat(amountLeft) { asmgen.out(" lda  P8ESTACK_HI+1,x |  asl  a  |  ror  P8ESTACK_HI+1,x |  ror  P8ESTACK_LO+1,x") }
-                            else
-                                asmgen.out(" jsr  math.shift_right_w_$amountLeft")
-                        }
-                        else -> throw AssemblyError("weird type")
-                    }
-                    return true
-                }
-            }
-            "<<" -> {
-                val amount = right.asConstInteger()
-                if(amount!=null) {
-                    translateExpressionInternal(left)
-                    if (leftDt in ByteDatatypes) {
-                        if (amount <= 2)
-                            repeat(amount) { asmgen.out("  asl  P8ESTACK_LO+1,x") }
-                        else {
-                            asmgen.out("  lda  P8ESTACK_LO+1,x")
-                            repeat(amount) { asmgen.out("  asl  a") }
-                            asmgen.out("  sta  P8ESTACK_LO+1,x")
-                        }
-                    } else {
-                        var amountLeft = amount
-                        while (amountLeft >= 7) {
-                            asmgen.out(" jsr  math.shift_left_w_7")
-                            amountLeft -= 7
-                        }
-                        if (amountLeft in 0..2)
-                            repeat(amountLeft) { asmgen.out("  asl  P8ESTACK_LO+1,x |  rol  P8ESTACK_HI+1,x") }
-                        else
-                            asmgen.out(" jsr  math.shift_left_w_$amountLeft")
-                    }
-                    return true
-                }
-            }
-            "*" -> {
-                if(leftDt in IntegerDatatypes && rightDt in IntegerDatatypes) {
-                    val leftVar = left as? PtIdentifier
-                    val rightVar = right as? PtIdentifier
-                    if(leftVar!=null && rightVar!=null && leftVar==rightVar) {
-                        translateSquared(leftVar, leftDt)
-                        return true
-                    }
-                }
-
-                val value = right as? PtNumber
-                if(value!=null) {
-                    if(rightDt in IntegerDatatypes) {
-                        val amount = value.number.toInt()
-                        if(amount==2) {
-                            // optimize x*2 common case
-                            translateExpressionInternal(left)
-                            if(leftDt in ByteDatatypes) {
-                                asmgen.out("  asl  P8ESTACK_LO+1,x")
-                            } else {
-                                asmgen.out("  asl  P8ESTACK_LO+1,x |  rol  P8ESTACK_HI+1,x")
-                            }
-                            return true
-                        }
-                        when(rightDt) {
-                            DataType.UBYTE -> {
-                                if(amount in asmgen.optimizedByteMultiplications) {
-                                    translateExpressionInternal(left)
-                                    asmgen.out(" jsr  math.stack_mul_byte_$amount")
-                                    return true
-                                }
-                            }
-                            DataType.BYTE -> {
-                                if(amount in asmgen.optimizedByteMultiplications) {
-                                    translateExpressionInternal(left)
-                                    asmgen.out(" jsr  math.stack_mul_byte_$amount")
-                                    return true
-                                }
-                                if(amount.absoluteValue in asmgen.optimizedByteMultiplications) {
-                                    translateExpressionInternal(left)
-                                    asmgen.out(" jsr  prog8_lib.neg_b |  jsr  math.stack_mul_byte_${amount.absoluteValue}")
-                                    return true
-                                }
-                            }
-                            DataType.UWORD -> {
-                                if(amount in asmgen.optimizedWordMultiplications) {
-                                    translateExpressionInternal(left)
-                                    asmgen.out(" jsr  math.stack_mul_word_$amount")
-                                    return true
-                                }
-                            }
-                            DataType.WORD -> {
-                                if(amount in asmgen.optimizedWordMultiplications) {
-                                    translateExpressionInternal(left)
-                                    asmgen.out(" jsr  math.stack_mul_word_$amount")
-                                    return true
-                                }
-                                if(amount.absoluteValue in asmgen.optimizedWordMultiplications) {
-                                    translateExpressionInternal(left)
-                                    asmgen.out(" jsr  prog8_lib.neg_w |  jsr  math.stack_mul_word_${amount.absoluteValue}")
-                                    return true
-                                }
-                            }
-                            else -> {}
-                        }
-                    }
-                }
-            }
-            "/" -> {
-                if(leftDt in IntegerDatatypes && rightDt in IntegerDatatypes) {
-                    val rightVal = right.asConstInteger()
-                    if(rightVal!=null && rightVal==2) {
-                        translateExpressionInternal(left)
-                        when (leftDt) {
-                            DataType.UBYTE -> {
-                                asmgen.out("  lsr  P8ESTACK_LO+1,x")
-                            }
-                            DataType.UWORD -> {
-                                asmgen.out("  lsr  P8ESTACK_HI+1,x |  ror  P8ESTACK_LO+1,x")
-                            }
-                            DataType.BYTE -> {
-                                // signed divide using shift needs adjusting of negative value to get correct rounding towards zero
-                                asmgen.out("""
-                                    lda  P8ESTACK_LO+1,x
-                                    bpl  +
-                                    inc  P8ESTACK_LO+1,x
-                                    lda  P8ESTACK_LO+1,x
-+                                   asl  a
-                                    ror  P8ESTACK_LO+1,x""")
-                            }
-                            DataType.WORD -> {
-                                // signed divide using shift needs adjusting of negative value to get correct rounding towards zero
-                                asmgen.out("""
-                                    lda  P8ESTACK_HI+1,x
-                                    bpl  ++
-                                    inc  P8ESTACK_LO+1,x
-                                    bne  +
-                                    inc  P8ESTACK_HI+1,x
-+                                   lda  P8ESTACK_HI+1,x
-+                                   asl  a
-                                    ror  P8ESTACK_HI+1,x
-                                    ror  P8ESTACK_LO+1,x""")
-                            }
-                            else -> throw AssemblyError("weird dt")
-                        }
-                        return true
-                    }
-                }
-            }
-        }
-
-        return false
-    }
-
-    private fun translateComparisonWithZero(expr: PtExpression, dt: DataType, operator: String) {
-        if(expr.isSimple()) {
-            if(operator=="!=") {
-                when (dt) {
-                    in ByteDatatypes -> {
-                        asmgen.assignExpressionToRegister(expr, RegisterOrPair.A, dt == DataType.BYTE)
-                        asmgen.out("""
-                            beq  +
-                            lda  #1
-+                           sta  P8ESTACK_LO,x
-                            dex""")
-                        return
-                    }
-                    in WordDatatypes -> {
-                        asmgen.assignExpressionToRegister(expr, RegisterOrPair.AY, dt == DataType.WORD)
-                        asmgen.out("""
-                            sty  P8ZP_SCRATCH_B1 
-                            ora  P8ZP_SCRATCH_B1
-                            beq  +
-                            lda  #1
-+                           sta  P8ESTACK_LO,x
-                            dex""")
-                        return
-                    }
-                    DataType.FLOAT -> {
-                        asmgen.assignExpressionToRegister(expr, RegisterOrPair.FAC1, true)
-                        asmgen.out("""
-                            jsr  floats.SIGN
-                            sta  P8ESTACK_LO,x
-                            dex""")
-                        return
-                    }
-                    else -> {}
-                }
-            }
-            /* operator == is not worth it to special case, the code is mostly larger */
-        }
-        translateExpressionInternal(expr)
-        when(operator) {
-            "==" -> {
-                when(dt) {
-                    DataType.UBYTE, DataType.BYTE -> asmgen.out("  jsr  prog8_lib.equalzero_b")
-                    DataType.UWORD, DataType.WORD -> asmgen.out("  jsr  prog8_lib.equalzero_w")
-                    DataType.FLOAT -> asmgen.out("  jsr  floats.equal_zero")
-                    else -> throw AssemblyError("wrong dt")
-                }
-            }
-            "!=" -> {
-                when(dt) {
-                    DataType.UBYTE, DataType.BYTE -> asmgen.out("  jsr  prog8_lib.notequalzero_b")
-                    DataType.UWORD, DataType.WORD -> asmgen.out("  jsr  prog8_lib.notequalzero_w")
-                    DataType.FLOAT -> asmgen.out("  jsr  floats.notequal_zero")
-                    else -> throw AssemblyError("wrong dt")
-                }
-            }
-            "<" -> {
-                if(dt==DataType.UBYTE || dt==DataType.UWORD)
-                    return translateExpressionInternal(PtNumber.fromBoolean(false, expr.position))
-                when(dt) {
-                    DataType.BYTE -> asmgen.out("  jsr  prog8_lib.lesszero_b")
-                    DataType.WORD -> asmgen.out("  jsr  prog8_lib.lesszero_w")
-                    DataType.FLOAT -> asmgen.out("  jsr  floats.less_zero")
-                    else -> throw AssemblyError("wrong dt")
-                }
-            }
-            ">" -> {
-                when(dt) {
-                    DataType.UBYTE -> asmgen.out("  jsr  prog8_lib.greaterzero_ub")
-                    DataType.BYTE -> asmgen.out("  jsr  prog8_lib.greaterzero_sb")
-                    DataType.UWORD -> asmgen.out("  jsr  prog8_lib.greaterzero_uw")
-                    DataType.WORD -> asmgen.out("  jsr  prog8_lib.greaterzero_sw")
-                    DataType.FLOAT -> asmgen.out("  jsr  floats.greater_zero")
-                    else -> throw AssemblyError("wrong dt")
-                }
-            }
-            "<=" -> {
-                when(dt) {
-                    DataType.UBYTE -> asmgen.out("  jsr  prog8_lib.equalzero_b")
-                    DataType.BYTE -> asmgen.out("  jsr  prog8_lib.lessequalzero_sb")
-                    DataType.UWORD -> asmgen.out("  jsr  prog8_lib.equalzero_w")
-                    DataType.WORD -> asmgen.out("  jsr  prog8_lib.lessequalzero_sw")
-                    DataType.FLOAT -> asmgen.out("  jsr  floats.lessequal_zero")
-                    else -> throw AssemblyError("wrong dt")
-                }
-            }
-            ">=" -> {
-                if(dt==DataType.UBYTE || dt==DataType.UWORD)
-                    return translateExpressionInternal(PtNumber.fromBoolean(true, expr.position))
-                when(dt) {
-                    DataType.BYTE -> asmgen.out("  jsr  prog8_lib.greaterequalzero_sb")
-                    DataType.WORD -> asmgen.out("  jsr  prog8_lib.greaterequalzero_sw")
-                    DataType.FLOAT -> asmgen.out("  jsr  floats.greaterequal_zero")
-                    else -> throw AssemblyError("wrong dt")
-                }
-            }
-            else -> throw AssemblyError("invalid comparison operator")
-        }
-    }
-
-    private fun translateSquared(variable: PtIdentifier, dt: DataType) {
-        val asmVar = asmgen.asmVariableName(variable)
-        when(dt) {
-            DataType.BYTE, DataType.UBYTE -> {
-                asmgen.out("  lda  $asmVar")
-                asmgen.signExtendAYlsb(dt)
-                asmgen.out("  jsr  math.square")
-            }
-            DataType.UWORD, DataType.WORD -> {
-                asmgen.out("  lda  $asmVar |  ldy  $asmVar+1 |  jsr  math.square")
-            }
-            else -> throw AssemblyError("require integer dt for square")
-        }
-        asmgen.out("  sta  P8ESTACK_LO,x |  tya |  sta  P8ESTACK_HI,x |  dex")
-    }
-
-    private fun translateExpression(expr: PtPrefix) {
-        translateExpressionInternal(expr.value)
-        when(expr.operator) {
-            "+" -> {}
-            "-" -> {
-                when(expr.type) {
-                    in ByteDatatypes -> asmgen.out("  jsr  prog8_lib.neg_b")
-                    in WordDatatypes -> asmgen.out("  jsr  prog8_lib.neg_w")
-                    DataType.FLOAT -> asmgen.out("  jsr  floats.neg_f")
-                    else -> throw AssemblyError("weird type")
-                }
-            }
-            "~" -> {
-                when(expr.type) {
-                    in ByteDatatypes ->
-                        asmgen.out("""
-                            lda  P8ESTACK_LO+1,x
-                            eor  #255
-                            sta  P8ESTACK_LO+1,x
-                            """)
-                    in WordDatatypes -> asmgen.out("  jsr  prog8_lib.inv_word")
-                    else -> throw AssemblyError("weird type")
-                }
-            }
-            else -> throw AssemblyError("invalid prefix operator ${expr.operator}")
-        }
-    }
-
-    private fun translateExpression(arrayExpr: PtArrayIndexer) {
-        val elementDt = arrayExpr.type
-        val arrayVarName = asmgen.asmVariableName(arrayExpr.variable)
-
-        if(arrayExpr.variable.type==DataType.UWORD) {
-            // indexing a pointer var instead of a real array or string
-            if(elementDt !in ByteDatatypes)
-                throw AssemblyError("non-array var indexing requires bytes dt")
-            if(arrayExpr.index.type != DataType.UBYTE)
-                throw AssemblyError("non-array var indexing requires bytes index")
-            asmgen.loadScaledArrayIndexIntoRegister(arrayExpr, elementDt, CpuRegister.Y)
-            if(asmgen.isZpVar(arrayExpr.variable)) {
-                asmgen.out("  lda  ($arrayVarName),y")
-            } else {
-                asmgen.out("  lda  $arrayVarName |  sta  P8ZP_SCRATCH_W1 |  lda  $arrayVarName+1 |  sta  P8ZP_SCRATCH_W1+1")
-                asmgen.out("  lda  (P8ZP_SCRATCH_W1),y")
-            }
-            asmgen.out("  sta  P8ESTACK_LO,x |  dex")
-            return
-        }
-
-        if(arrayExpr.splitWords)
-            TODO("split words expression ${arrayExpr.position}")
-
-        val constIndexNum = arrayExpr.index.asConstInteger()
-        if(constIndexNum!=null) {
-            val indexValue = constIndexNum * program.memsizer.memorySize(elementDt)
-            when(elementDt) {
-                in ByteDatatypes -> {
-                    asmgen.out("  lda  $arrayVarName+$indexValue |  sta  P8ESTACK_LO,x |  dex")
-                }
-                in WordDatatypes -> {
-                    asmgen.out("  lda  $arrayVarName+$indexValue |  sta  P8ESTACK_LO,x |  lda  $arrayVarName+$indexValue+1 |  sta  P8ESTACK_HI,x |  dex")
-                }
-                DataType.FLOAT -> {
-                    asmgen.out("  lda  #<($arrayVarName+$indexValue) |  ldy  #>($arrayVarName+$indexValue) |  jsr  floats.push_float")
-                }
-                else -> throw AssemblyError("weird element type")
-            }
-        } else {
-            when(elementDt) {
-                in ByteDatatypes -> {
-                    asmgen.loadScaledArrayIndexIntoRegister(arrayExpr, elementDt, CpuRegister.Y)
-                    asmgen.out(" lda  $arrayVarName,y |  sta  P8ESTACK_LO,x |  dex")
-                }
-                in WordDatatypes -> {
-                    asmgen.loadScaledArrayIndexIntoRegister(arrayExpr, elementDt, CpuRegister.Y)
-                    asmgen.out(" lda  $arrayVarName,y |  sta  P8ESTACK_LO,x |  lda  $arrayVarName+1,y |  sta  P8ESTACK_HI,x |  dex")
-                }
-                DataType.FLOAT -> {
-                    asmgen.loadScaledArrayIndexIntoRegister(arrayExpr, elementDt, CpuRegister.A)
-                    asmgen.out("""
-                        ldy  #>$arrayVarName
-                        clc
-                        adc  #<$arrayVarName
-                        bcc  +
-                        iny
-+                       jsr  floats.push_float""")
-                }
-                else -> throw AssemblyError("weird dt")
-            }
-
-        }
-    }
-
-    private fun translateBinaryOperatorBytes(operator: String, types: DataType) {
-        when(operator) {
-            "*" -> asmgen.out("  jsr  prog8_lib.mul_byte")  //  the optimized routines should have been checked earlier
-            "/" -> asmgen.out(if(types==DataType.UBYTE) "  jsr  prog8_lib.idiv_ub" else "  jsr  prog8_lib.idiv_b")
-            "%" -> {
-                if(types==DataType.BYTE)
-                    throw AssemblyError("remainder of signed integers is not properly defined/implemented, use unsigned instead")
-                asmgen.out("  jsr prog8_lib.remainder_ub")
-            }
-            "+" -> asmgen.out("""
-                lda  P8ESTACK_LO+2,x
-                clc
-                adc  P8ESTACK_LO+1,x
-                inx
-                sta  P8ESTACK_LO+1,x
-                """)
-            "-" -> asmgen.out("""
-                lda  P8ESTACK_LO+2,x
-                sec
-                sbc  P8ESTACK_LO+1,x
-                inx
-                sta  P8ESTACK_LO+1,x
-                """)
-            "<<" -> asmgen.out("  jsr  prog8_lib.shiftleft_b")
-            ">>" -> asmgen.out("  jsr  prog8_lib.shiftright_b")
-            "<" -> asmgen.out(if(types==DataType.UBYTE) "  jsr  prog8_lib.less_ub" else "  jsr  prog8_lib.less_b")
-            ">" -> asmgen.out(if(types==DataType.UBYTE) "  jsr  prog8_lib.greater_ub" else "  jsr  prog8_lib.greater_b")
-            "<=" -> asmgen.out(if(types==DataType.UBYTE) "  jsr  prog8_lib.lesseq_ub" else "  jsr  prog8_lib.lesseq_b")
-            ">=" -> asmgen.out(if(types==DataType.UBYTE) "  jsr  prog8_lib.greatereq_ub" else "  jsr  prog8_lib.greatereq_b")
-            "==" -> asmgen.out("  jsr  prog8_lib.equal_b")
-            "!=" -> asmgen.out("  jsr  prog8_lib.notequal_b")
-            "&" -> asmgen.out("  jsr  prog8_lib.bitand_b")
-            "^" -> asmgen.out("  jsr  prog8_lib.bitxor_b")
-            "|" -> asmgen.out("  jsr  prog8_lib.bitor_b")
-            else -> throw AssemblyError("invalid operator $operator")
-        }
-    }
-
-    private fun translateBinaryOperatorWords(operator: String, dt: DataType) {
-        when(operator) {
-            "*" -> asmgen.out("  jsr  prog8_lib.mul_word")
-            "/" -> asmgen.out(if(dt==DataType.UWORD) "  jsr  prog8_lib.idiv_uw" else "  jsr  prog8_lib.idiv_w")
-            "%" -> {
-                if(dt==DataType.WORD)
-                    throw AssemblyError("remainder of signed integers is not properly defined/implemented, use unsigned instead")
-                asmgen.out("  jsr prog8_lib.remainder_uw")
-            }
-            "+" -> asmgen.out("  jsr  prog8_lib.add_w")
-            "-" -> asmgen.out("  jsr  prog8_lib.sub_w")
-            "<<" -> asmgen.out("  jsr  math.shift_left_w")
-            ">>" -> {
-                if(dt==DataType.UWORD)
-                    asmgen.out("  jsr  math.shift_right_uw")
-                else
-                    asmgen.out("  jsr  math.shift_right_w")
-            }
-            "<" -> asmgen.out(if(dt==DataType.UWORD) "  jsr  prog8_lib.less_uw" else "  jsr  prog8_lib.less_w")
-            ">" -> asmgen.out(if(dt==DataType.UWORD) "  jsr  prog8_lib.greater_uw" else "  jsr  prog8_lib.greater_w")
-            "<=" -> asmgen.out(if(dt==DataType.UWORD) "  jsr  prog8_lib.lesseq_uw" else "  jsr  prog8_lib.lesseq_w")
-            ">=" -> asmgen.out(if(dt==DataType.UWORD) "  jsr  prog8_lib.greatereq_uw" else "  jsr  prog8_lib.greatereq_w")
-            "==" -> asmgen.out("  jsr  prog8_lib.equal_w")
-            "!=" -> asmgen.out("  jsr  prog8_lib.notequal_w")
-            "&" -> asmgen.out("  jsr  prog8_lib.bitand_w")
-            "^" -> asmgen.out("  jsr  prog8_lib.bitxor_w")
-            "|" -> asmgen.out("  jsr  prog8_lib.bitor_w")
-            else -> throw AssemblyError("invalid operator $operator")
-        }
-    }
-
-    private fun translateBinaryOperatorFloats(operator: String) {
-        when(operator) {
-            "*" -> asmgen.out("  jsr  floats.mul_f")
-            "/" -> asmgen.out("  jsr  floats.div_f")
-            "+" -> asmgen.out("  jsr  floats.add_f")
-            "-" -> asmgen.out("  jsr  floats.sub_f")
-            "<" -> asmgen.out("  jsr  floats.less_f")
-            ">" -> asmgen.out("  jsr  floats.greater_f")
-            "<=" -> asmgen.out("  jsr  floats.lesseq_f")
-            ">=" -> asmgen.out("  jsr  floats.greatereq_f")
-            "==" -> asmgen.out("  jsr  floats.equal_f")
-            "!=" -> asmgen.out("  jsr  floats.notequal_f")
-            "%", "<<", ">>", "&", "^", "|" -> throw AssemblyError("requires integer datatype")
-            else -> throw AssemblyError("invalid operator $operator")
-        }
-    }
-
-    private fun translateCompareStrings(s1: PtExpression, operator: String, s2: PtExpression) {
-        asmgen.assignExpressionToVariable(s1, "prog8_lib.strcmp_expression._arg_s1", DataType.UWORD)
-        asmgen.assignExpressionToVariable(s2, "prog8_lib.strcmp_expression._arg_s2", DataType.UWORD)
-        asmgen.out(" jsr  prog8_lib.strcmp_expression")    // result  of compare is in A
-        compareStringsProcessResultInA(operator)
-    }
-
-    private fun compareStringsProcessResultInA(operator: String) {
-        when(operator) {
-            "==" -> asmgen.out(" and  #1 |  eor  #1 |  sta  P8ESTACK_LO,x")
-            "!=" -> asmgen.out(" and  #1 |  sta  P8ESTACK_LO,x")
-            "<=" -> asmgen.out("""
-                bpl  +
-                lda  #1
-                bne  ++
-+               lda  #0
-+               sta  P8ESTACK_LO,x""")
-            ">=" -> asmgen.out("""
-                bmi  +
-                lda  #1
-                bne  ++
-+               lda  #0
-+               sta  P8ESTACK_LO,x""")
-            "<" -> asmgen.out("""
-                bmi  +
-                lda  #0
-                beq  ++
-+               lda  #1
-+               sta  P8ESTACK_LO,x""")
-            ">" -> asmgen.out("""
-                bpl  +
-                lda  #0
-                beq  ++
-+               lda  #1
-+               sta  P8ESTACK_LO,x""")
-        }
-        asmgen.out("  dex")
-    }
-}

codeGenCpu6502/src/prog8/codegen/cpu6502/Extensions.kt

@@ -6,26 +6,6 @@ import prog8.code.ast.PtSub
 import prog8.code.core.*
 
 
-internal fun IPtSubroutine.regXasResult(): Boolean =
-    (this is PtAsmSub) && this.returns.any { it.first.registerOrPair in arrayOf(RegisterOrPair.X, RegisterOrPair.AX, RegisterOrPair.XY) }
-
-internal fun IPtSubroutine.shouldSaveX(): Boolean =
-    this.regXasResult() || (this is PtAsmSub && (CpuRegister.X in this.clobbers || regXasParam()))
-
-internal fun PtAsmSub.regXasParam(): Boolean =
-    parameters.any { it.first.registerOrPair in arrayOf(RegisterOrPair.X, RegisterOrPair.AX, RegisterOrPair.XY) }
-
-internal class KeepAresult(val saveOnEntry: Boolean, val saveOnReturn: Boolean)
-
-internal fun PtAsmSub.shouldKeepA(): KeepAresult {
-    // determine if A's value should be kept when preparing for calling the subroutine, and when returning from it
-
-    // it seems that we never have to save A when calling? will be loaded correctly after setup.
-    // but on return it depends on wether the routine returns something in A.
-    val saveAonReturn = returns.any { it.first.registerOrPair==RegisterOrPair.A || it.first.registerOrPair==RegisterOrPair.AY || it.first.registerOrPair==RegisterOrPair.AX }
-    return KeepAresult(false, saveAonReturn)
-}
-
 internal fun IPtSubroutine.returnsWhatWhere(): List<Pair<RegisterOrStatusflag, DataType>> {
     when(this) {
         is PtAsmSub -> {

codeGenCpu6502/src/prog8/codegen/cpu6502/ForLoopsAsmGen.kt

@@ -5,9 +5,10 @@ import prog8.code.ast.*
 import prog8.code.core.*
 import kotlin.math.absoluteValue
 
-internal class ForLoopsAsmGen(private val program: PtProgram,
-                              private val asmgen: AsmGen6502Internal,
-                              private val zeropage: Zeropage) {
+internal class ForLoopsAsmGen(
+    private val asmgen: AsmGen6502Internal,
+    private val zeropage: Zeropage
+) {
 
     internal fun translate(stmt: PtForLoop) {
         val iterableDt = stmt.iterable.type
@@ -249,33 +250,16 @@ $endLabel""")
 $loopLabel""")
                         asmgen.translate(stmt.statements)
 
-                        if(iterableDt==DataType.ARRAY_UW) {
                             asmgen.out("""
                 lda  $varname
                 sec
                 sbc  #<${stepsize.absoluteValue}
                 sta  $varname
+                tax
                 lda  $varname+1
                 sbc  #>${stepsize.absoluteValue}
                 sta  $varname+1
-$modifiedLabel  cmp  #0    ; modified                
-                bcc  $endLabel
-                bne  $loopLabel
-                lda  $varname
-$modifiedLabel2 cmp  #0    ; modified 
-                bcs  $loopLabel
-$endLabel""")
-                        } else {
-                            asmgen.out("""
-                lda  $varname
-                sec
-                sbc  #<${stepsize.absoluteValue}
-                sta  $varname
-                pha
-                lda  $varname+1
-                sbc  #>${stepsize.absoluteValue}
-                sta  $varname+1
-                pla
+                txa
 $modifiedLabel2 cmp  #0    ; modified 
                 lda  $varname+1
 $modifiedLabel  sbc  #0    ; modified
@@ -283,7 +267,6 @@ $modifiedLabel  sbc  #0    ; modified
                 eor  #$80
 +               bpl  $loopLabel                
 $endLabel""")
-                        }
                     }
                 }
             }
@@ -400,7 +383,7 @@ $loopLabel          sty  $indexVar
                     // allocate index var on ZP if possible
                     val result = zeropage.allocate(indexVar, DataType.UBYTE, null, stmt.position, asmgen.errors)
                     result.fold(
-                        success = { (address,_)-> asmgen.out("""$indexVar = $address  ; auto zp UBYTE""") },
+                        success = { (address, _, _)-> asmgen.out("""$indexVar = $address  ; auto zp UBYTE""") },
                         failure = { asmgen.out("$indexVar    .byte  0") }
                     )
                 } else {

codeGenCpu6502/src/prog8/codegen/cpu6502/FunctionCallAsmGen.kt

@@ -11,42 +11,10 @@ import prog8.codegen.cpu6502.assignment.TargetStorageKind
 internal class FunctionCallAsmGen(private val program: PtProgram, private val asmgen: AsmGen6502Internal) {
 
     internal fun translateFunctionCallStatement(stmt: PtFunctionCall) {
-        saveXbeforeCall(stmt)
         translateFunctionCall(stmt)
-        restoreXafterCall(stmt)
         // just ignore any result values from the function call.
     }
 
-    internal fun saveXbeforeCall(stmt: PtFunctionCall) {
-        val symbol = asmgen.symbolTable.lookup(stmt.name)
-        val sub = symbol!!.astNode as IPtSubroutine
-        if(sub.shouldSaveX()) {
-            if(sub is PtAsmSub) {
-                val regSaveOnStack = sub.address == null       // rom-routines don't require registers to be saved on stack, normal subroutines do because they can contain nested calls
-                if (regSaveOnStack)
-                    asmgen.saveRegisterStack(CpuRegister.X, sub.shouldKeepA().saveOnEntry)
-                else
-                    asmgen.saveRegisterLocal(CpuRegister.X, stmt.definingISub()!!)
-            } else
-                asmgen.saveRegisterLocal(CpuRegister.X, stmt.definingISub()!!)
-        }
-    }
-
-    internal fun restoreXafterCall(stmt: PtFunctionCall) {
-        val symbol = asmgen.symbolTable.lookup(stmt.name)
-        val sub = symbol!!.astNode as IPtSubroutine
-        if(sub.shouldSaveX()) {
-            if(sub is PtAsmSub) {
-                val regSaveOnStack = sub.address == null       // rom-routines don't require registers to be saved on stack, normal subroutines do because they can contain nested calls
-                if (regSaveOnStack)
-                    asmgen.restoreRegisterStack(CpuRegister.X, sub.shouldKeepA().saveOnReturn)
-                else
-                    asmgen.restoreRegisterLocal(CpuRegister.X)
-            } else
-                asmgen.restoreRegisterLocal(CpuRegister.X)
-        }
-    }
-
     internal fun optimizeIntArgsViaRegisters(sub: PtSub) =
         (sub.parameters.size==1 && sub.parameters[0].type in IntegerDatatypes)
                 || (sub.parameters.size==2 && sub.parameters[0].type in ByteDatatypes && sub.parameters[1].type in ByteDatatypes)
@@ -63,8 +31,8 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
 
         if(sub is PtAsmSub) {
             argumentsViaRegisters(sub, call)
-            if (sub.inline && asmgen.options.optimize) {
-                // inline the subroutine.
+            if (sub.inline) {
+                // inline the subroutine. (regardless of optimization settings!)
                 // we do this by copying the subroutine's statements at the call site.
                 // NOTE: *if* there is a return statement, it will be the only one, and the very last statement of the subroutine
                 // (this condition has been enforced by an ast check earlier)
@@ -174,12 +142,12 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
                     is PtIdentifier -> {
                         val sourceName = asmgen.asmVariableName(value)
                         asmgen.out("""
-                            pha
+                            tax
                             clc
                             lda  $sourceName
                             beq  +
                             sec  
-+                           pla""")
++                           txa""")
                     }
                     else -> {
                         asmgen.assignExpressionToRegister(value, RegisterOrPair.A)

codeGenCpu6502/src/prog8/codegen/cpu6502/PostIncrDecrAsmGen.kt

@@ -13,7 +13,6 @@ internal class PostIncrDecrAsmGen(private val program: PtProgram, private val as
         val targetIdent = stmt.target.identifier
         val targetMemory = stmt.target.memory
         val targetArrayIdx = stmt.target.array
-        val scope = stmt.definingISub()
         when {
             targetIdent!=null -> {
                 val what = asmgen.asmVariableName(targetIdent)
@@ -76,7 +75,6 @@ internal class PostIncrDecrAsmGen(private val program: PtProgram, private val as
         dec  ${asmArrayvarname}_msb+$constIndex
 +       dec  ${asmArrayvarname}_lsb+$constIndex""")
                     } else {
-                        asmgen.saveRegisterLocal(CpuRegister.X, scope!!)
                         asmgen.loadScaledArrayIndexIntoRegister(targetArrayIdx, elementDt, CpuRegister.X)
                         if(incr)
                             asmgen.out(" inc  ${asmArrayvarname}_lsb,x |  bne  + |  inc  ${asmArrayvarname}_msb,x |+")
@@ -86,14 +84,30 @@ internal class PostIncrDecrAsmGen(private val program: PtProgram, private val as
         bne  +
         dec  ${asmArrayvarname}_msb,x
 +       dec  ${asmArrayvarname}_lsb,x""")
-                        asmgen.restoreRegisterLocal(CpuRegister.X)
                     }
                     return
                 }
                 if(constIndex!=null) {
                     val indexValue = constIndex * program.memsizer.memorySize(elementDt)
                     when(elementDt) {
-                        in ByteDatatypes -> asmgen.out(if (incr) "  inc  $asmArrayvarname+$indexValue" else "  dec  $asmArrayvarname+$indexValue")
+                        in ByteDatatypes -> {
+                            if(targetArrayIdx.usesPointerVariable) {
+                                asmgen.out("""
+                                    lda  $asmArrayvarname
+                                    clc
+                                    adc  #$indexValue
+                                    sta  (+) +1
+                                    lda  $asmArrayvarname+1
+                                    adc  #0
+                                    sta  (+) +2""")
+                                if(incr)
+                                    asmgen.out("+\tinc  ${'$'}ffff\t; modified")
+                                else
+                                    asmgen.out("+\tdec  ${'$'}ffff\t; modified")
+                            } else {
+                                asmgen.out(if (incr) "  inc  $asmArrayvarname+$indexValue" else "  dec  $asmArrayvarname+$indexValue")
+                            }
+                        }
                         in WordDatatypes -> {
                             if(incr)
                                 asmgen.out(" inc  $asmArrayvarname+$indexValue |  bne  + |  inc  $asmArrayvarname+$indexValue+1 |+")
@@ -113,11 +127,25 @@ internal class PostIncrDecrAsmGen(private val program: PtProgram, private val as
                 }
                 else
                 {
-                    asmgen.saveRegisterLocal(CpuRegister.X, scope!!)
                     asmgen.loadScaledArrayIndexIntoRegister(targetArrayIdx, elementDt, CpuRegister.X)
                     when(elementDt) {
                         in ByteDatatypes -> {
-                            asmgen.out(if(incr) "  inc  $asmArrayvarname,x" else "  dec  $asmArrayvarname,x")
+                            if(targetArrayIdx.usesPointerVariable) {
+                                asmgen.out("""
+                                    txa
+                                    clc
+                                    adc  $asmArrayvarname
+                                    sta  (+) +1
+                                    lda  $asmArrayvarname+1
+                                    adc  #0
+                                    sta  (+) +2""")
+                                if(incr)
+                                    asmgen.out("+\tinc  ${'$'}ffff\t; modified")
+                                else
+                                    asmgen.out("+\tdec  ${'$'}ffff\t; modified")
+                            } else {
+                                asmgen.out(if (incr) "  inc  $asmArrayvarname,x" else "  dec  $asmArrayvarname,x")
+                            }
                         }
                         in WordDatatypes -> {
                             if(incr)
@@ -127,8 +155,7 @@ internal class PostIncrDecrAsmGen(private val program: PtProgram, private val as
         lda  $asmArrayvarname,x
         bne  +
         dec  $asmArrayvarname+1,x
-+       dec  $asmArrayvarname,x 
-""")
++       dec  $asmArrayvarname,x""")
                         }
                         DataType.FLOAT -> {
                             asmgen.out("""
@@ -141,7 +168,6 @@ internal class PostIncrDecrAsmGen(private val program: PtProgram, private val as
                         }
                         else -> throw AssemblyError("weird array elt dt")
                     }
-                    asmgen.restoreRegisterLocal(CpuRegister.X)
                 }
             }
         }

codeGenCpu6502/src/prog8/codegen/cpu6502/ProgramAndVarsGen.kt

@@ -33,8 +33,9 @@ internal class ProgramAndVarsGen(
     internal fun generate() {
         header()
         val allBlocks = program.allBlocks()
-        if(allBlocks.first().name != "main")
-            throw AssemblyError("first block should be 'main'")
+
+        if(allBlocks.first().name != "p8_main" && allBlocks.first().name != "main")
+            throw AssemblyError("first block should be 'main' or 'p8_main'")
 
         if(errors.noErrors())  {
             program.allBlocks().forEach { block2asm(it) }
@@ -74,8 +75,6 @@ internal class ProgramAndVarsGen(
         asmgen.out("P8ZP_SCRATCH_W1 = ${zp.SCRATCH_W1}    ; word")
         asmgen.out("P8ZP_SCRATCH_W2 = ${zp.SCRATCH_W2}    ; word")
         asmgen.out(".weak")   // hack to allow user to override the following two with command line redefinition (however, just use '-esa' command line option instead!)
-        asmgen.out("P8ESTACK_LO = ${compTarget.machine.ESTACK_LO.toHex()}")
-        asmgen.out("P8ESTACK_HI = ${compTarget.machine.ESTACK_HI.toHex()}")
         asmgen.out(".endweak")
 
         if(options.symbolDefs.isNotEmpty()) {
@@ -139,24 +138,24 @@ internal class ProgramAndVarsGen(
             "cx16" -> {
                 if(options.floats)
                     asmgen.out("  lda  #4 |  sta  $01")    // to use floats, make sure Basic rom is banked in
-                asmgen.out("  jsr  main.start")
+                asmgen.out("  jsr  p8_main.p8_start")
                 asmgen.out("  jmp  sys.cleanup_at_exit")
             }
             "c64" -> {
-                asmgen.out("  jsr  main.start |  lda  #31 |  sta  $01")
+                asmgen.out("  jsr  p8_main.p8_start |  lda  #31 |  sta  $01")
                 if(!options.noSysInit)
                     asmgen.out("  jmp  sys.cleanup_at_exit")
                 else
                     asmgen.out("  rts")
             }
             "c128" -> {
-                asmgen.out("  jsr  main.start |  lda  #0 |  sta ${"$"}ff00")
+                asmgen.out("  jsr  p8_main.p8_start |  lda  #0 |  sta ${"$"}ff00")
                 if(!options.noSysInit)
                     asmgen.out("  jmp  sys.cleanup_at_exit")
                 else
                     asmgen.out("  rts")
             }
-            else -> asmgen.jmp("main.start")
+            else -> asmgen.jmp("p8_main.p8_start")
         }
     }
 
@@ -175,7 +174,8 @@ internal class ProgramAndVarsGen(
 
     private fun footer() {
         asmgen.out("; bss sections")
-        if(options.varsHigh) {
+        asmgen.out("PROG8_VARSHIGH_RAMBANK = ${options.varsHighBank ?: 1}")
+        if(options.varsHighBank!=null) {
             if(options.compTarget.machine.BSSHIGHRAM_START == 0u || options.compTarget.machine.BSSHIGHRAM_END==0u) {
                 throw AssemblyError("current compilation target hasn't got the high ram area properly defined")
             }
@@ -269,9 +269,7 @@ internal class ProgramAndVarsGen(
 
     internal fun translateAsmSubroutine(sub: PtAsmSub) {
         if(sub.inline) {
-            if(options.optimize) {
-                return
-            }
+            return      // subroutine gets inlined at call site.
         }
 
         asmgen.out("")
@@ -311,7 +309,9 @@ internal class ProgramAndVarsGen(
 
         asmgen.out("${sub.name}\t$asmStartScope")
 
-        val scope = symboltable.lookupOrElse(sub.scopedName) { throw AssemblyError("lookup") }
+        val scope = symboltable.lookupOrElse(sub.scopedName) {
+            throw AssemblyError("lookup ${sub.scopedName}")
+        }
         require(scope.type==StNodeType.SUBROUTINE)
         val varsInSubroutine = getVars(scope)
 
@@ -331,7 +331,7 @@ internal class ProgramAndVarsGen(
         asmsubs2asm(sub.children)
 
         // the main.start subroutine is the program's entrypoint and should perform some initialization logic
-        if(sub.name=="start" && sub.definingBlock()!!.name=="main")
+        if((sub.name=="start" || sub.name=="p8_start") && (sub.definingBlock()!!.name=="main" || sub.definingBlock()!!.name=="p8_main"))
             entrypointInitialization()
 
         if(functioncallAsmGen.optimizeIntArgsViaRegisters(sub)) {
@@ -369,12 +369,6 @@ internal class ProgramAndVarsGen(
                 else -> throw AssemblyError("weird dt for extravar $dt")
             }
         }
-        if(asmGenInfo.usedRegsaveA)      // will probably never occur
-            asmgen.out("prog8_regsaveA     .byte  ?")
-        if(asmGenInfo.usedRegsaveX)
-            asmgen.out("prog8_regsaveX     .byte  ?")
-        if(asmGenInfo.usedRegsaveY)
-            asmgen.out("prog8_regsaveY     .byte  ?")
         if(asmGenInfo.usedFloatEvalResultVar1)
             asmgen.out("$subroutineFloatEvalResultVar1    .fill  ${options.compTarget.machine.FLOAT_MEM_SIZE}")
         if(asmGenInfo.usedFloatEvalResultVar2)
@@ -434,12 +428,12 @@ internal class ProgramAndVarsGen(
                 asmgen.out("""
                     lda  #<${name}_init_value
                     ldy  #>${name}_init_value
-                    sta  cx16.r0L
-                    sty  cx16.r0H
+                    sta  cx16.r0
+                    sty  cx16.r0+1
                     lda  #<${name}
                     ldy  #>${name}
-                    sta  cx16.r1L
-                    sty  cx16.r1H
+                    sta  cx16.r1
+                    sty  cx16.r1+1
                     lda  #<$size
                     ldy  #>$size
                     jsr  sys.memcopy""")
@@ -458,7 +452,6 @@ internal class ProgramAndVarsGen(
         }
 
         asmgen.out("""+        
-                    ldx  #127       ; init estack ptr (half page)
                     clv
                     clc""")
     }
@@ -480,8 +473,8 @@ internal class ProgramAndVarsGen(
         val vars = allocator.zeropageVars.filter { it.value.dt==DataType.STR }
         for (variable in vars) {
             val scopedName = variable.key
-            val svar = symboltable.flat.getValue(scopedName) as StStaticVariable
-            if(svar.onetimeInitializationStringValue!=null)
+            val svar = symboltable.lookup(scopedName) as? StStaticVariable
+            if(svar?.onetimeInitializationStringValue!=null)
                 result.add(ZpStringWithInitial(scopedName, variable.value, svar.onetimeInitializationStringValue!!))
         }
         return result
@@ -492,8 +485,8 @@ internal class ProgramAndVarsGen(
         val vars = allocator.zeropageVars.filter { it.value.dt in ArrayDatatypes }
         for (variable in vars) {
             val scopedName = variable.key
-            val svar = symboltable.flat.getValue(scopedName) as StStaticVariable
-            if(svar.onetimeInitializationArrayValue!=null)
+            val svar = symboltable.lookup(scopedName) as? StStaticVariable
+            if(svar?.onetimeInitializationArrayValue!=null)
                 result.add(ZpArrayWithInitial(scopedName, variable.value, svar.onetimeInitializationArrayValue!!))
         }
         return result

codeGenCpu6502/src/prog8/codegen/cpu6502/assignment/AnyExprAsmGen.kt

@@ -0,0 +1,281 @@
+package prog8.codegen.cpu6502.assignment
+
+import prog8.code.ast.PtBinaryExpression
+import prog8.code.ast.PtExpression
+import prog8.code.core.*
+import prog8.code.target.C64Target
+import prog8.code.target.Cx16Target
+import prog8.codegen.cpu6502.AsmGen6502Internal
+
+//
+// This contains codegen for stack-based evaluation of binary expressions.
+// It uses the CPU stack so depth is limited.
+// It is called "as a last resort" if the optimized codegen path is unable
+// to come up with a special case of the expression.
+//
+internal class AnyExprAsmGen(
+    private val asmgen: AsmGen6502Internal
+) {
+    fun assignAnyExpressionUsingStack(expr: PtBinaryExpression, assign: AsmAssignment): Boolean {
+        when(expr.type) {
+            in ByteDatatypes -> {
+                if(expr.left.type in ByteDatatypes && expr.right.type in ByteDatatypes)
+                    return assignByteBinExpr(expr, assign)
+                if (expr.left.type in WordDatatypes && expr.right.type in WordDatatypes) {
+                    require(expr.operator in ComparisonOperators)
+                    TODO("words operands comparison -> byte")
+                }
+                if (expr.left.type==DataType.FLOAT && expr.right.type==DataType.FLOAT) {
+                    require(expr.operator in ComparisonOperators)
+                    return assignFloatBinExpr(expr, assign)
+                }
+                TODO("weird expr operand types")
+            }
+            in WordDatatypes -> {
+                require(expr.left.type in WordDatatypes && expr.right.type in WordDatatypes) {
+                    "both operands must be words"
+                }
+                return assignWordBinExpr(expr)
+            }
+            DataType.FLOAT -> {
+                require(expr.left.type==DataType.FLOAT && expr.right.type==DataType.FLOAT) {
+                    "both operands must be floats"
+                }
+                return assignFloatBinExpr(expr, assign)
+            }
+            else -> throw AssemblyError("weird expression type in assignment")
+        }
+    }
+
+    private fun assignWordBinExpr(expr: PtBinaryExpression): Boolean {
+        when(expr.operator) {
+            "+" -> {
+                TODO("word + at ${expr.position}")
+            }
+            "-" -> {
+                TODO("word - at ${expr.position}")
+            }
+            "*" -> {
+                TODO("word * at ${expr.position}")
+            }
+            "/" -> {
+                TODO("word / at ${expr.position}")
+            }
+            "<<" -> {
+                TODO("word << at ${expr.position}")
+            }
+            ">>" -> {
+                TODO("word >> at ${expr.position}")
+            }
+            "%" -> {
+                TODO("word % at ${expr.position}")
+            }
+            "&", "and" -> {
+                TODO("word and at ${expr.position}")
+            }
+            "|", "or" -> {
+                TODO("word or at ${expr.position}")
+            }
+            "^", "xor" -> {
+                TODO("word xor at ${expr.position}")
+            }
+            "==" -> {
+                TODO("word == at ${expr.position}")
+            }
+            "!=" -> {
+                TODO("word != at ${expr.position}")
+            }
+            "<" -> {
+                TODO("word < at ${expr.position}")
+            }
+            "<=" -> {
+                TODO("word <= at ${expr.position}")
+            }
+            ">" -> {
+                TODO("word > at ${expr.position}")
+            }
+            ">=" -> {
+                TODO("word >= at ${expr.position}")
+            }
+            else -> return false
+        }
+    }
+
+    private fun assignByteBinExpr(expr: PtBinaryExpression, assign: AsmAssignment): Boolean {
+        when(expr.operator) {
+            "+" -> {
+                asmgen.assignExpressionToRegister(expr.left, RegisterOrPair.A, false)
+                asmgen.out("  pha")
+                asmgen.assignExpressionToVariable(expr.right, "P8ZP_SCRATCH_B1", DataType.UBYTE)
+                asmgen.out("  pla |  clc |  adc  P8ZP_SCRATCH_B1")
+                asmgen.assignRegister(RegisterOrPair.A, assign.target)
+                return true
+            }
+            "-" -> {
+                asmgen.assignExpressionToRegister(expr.left, RegisterOrPair.A, false)
+                asmgen.out("  pha")
+                asmgen.assignExpressionToVariable(expr.right, "P8ZP_SCRATCH_B1", DataType.UBYTE)
+                asmgen.out("  pla |  sec |  sbc  P8ZP_SCRATCH_B1")
+                asmgen.assignRegister(RegisterOrPair.A, assign.target)
+                return true
+            }
+            "*" -> {
+                TODO("byte * at ${expr.position}")
+            }
+            "/" -> {
+                TODO("byte / at ${expr.position}")
+            }
+            "<<" -> {
+                TODO("byte << at ${expr.position}")
+            }
+            ">>" -> {
+                TODO("byte >> at ${expr.position}")
+            }
+            "%" -> {
+                TODO("byte % at ${expr.position}")
+            }
+            "&", "and" -> {
+                asmgen.assignExpressionToRegister(expr.left, RegisterOrPair.A, false)
+                asmgen.out("  pha")
+                asmgen.assignExpressionToVariable(expr.right, "P8ZP_SCRATCH_B1", DataType.UBYTE)
+                asmgen.out("  pla |  and  P8ZP_SCRATCH_B1")
+                asmgen.assignRegister(RegisterOrPair.A, assign.target)
+                return true
+            }
+            "|", "or" -> {
+                asmgen.assignExpressionToRegister(expr.left, RegisterOrPair.A, false)
+                asmgen.out("  pha")
+                asmgen.assignExpressionToVariable(expr.right, "P8ZP_SCRATCH_B1", DataType.UBYTE)
+                asmgen.out("  pla |  ora  P8ZP_SCRATCH_B1")
+                asmgen.assignRegister(RegisterOrPair.A, assign.target)
+                return true
+            }
+            "^", "xor" -> {
+                asmgen.assignExpressionToRegister(expr.left, RegisterOrPair.A, false)
+                asmgen.out("  pha")
+                asmgen.assignExpressionToVariable(expr.right, "P8ZP_SCRATCH_B1", DataType.UBYTE)
+                asmgen.out("  pla |  eor  P8ZP_SCRATCH_B1")
+                asmgen.assignRegister(RegisterOrPair.A, assign.target)
+                return true
+            }
+            "==" -> {
+                TODO("byte == at ${expr.position}")
+            }
+            "!=" -> {
+                TODO("byte != at ${expr.position}")
+            }
+            "<" -> {
+                TODO("byte < at ${expr.position}")
+            }
+            "<=" -> {
+                TODO("byte <= at ${expr.position}")
+            }
+            ">" -> {
+                TODO("byte > at ${expr.position}")
+            }
+            ">=" -> {
+                TODO("byte >= at ${expr.position}")
+            }
+            else -> return false
+        }
+    }
+
+    private fun assignFloatBinExpr(expr: PtBinaryExpression, assign: AsmAssignment): Boolean {
+        when(expr.operator) {
+            "+" -> {
+                assignFloatOperandsToFACandARG(expr.left, expr.right)
+                asmgen.out("  jsr  floats.FADDT")
+                asmgen.assignRegister(RegisterOrPair.FAC1, assign.target)
+                return true
+            }
+            "-" -> {
+                assignFloatOperandsToFACandARG(expr.right, expr.left)
+                asmgen.out("  jsr  floats.FSUBT")
+                asmgen.assignRegister(RegisterOrPair.FAC1, assign.target)
+                return true
+            }
+            "*" -> {
+                assignFloatOperandsToFACandARG(expr.left, expr.right)
+                asmgen.out("  jsr  floats.FMULTT")
+                asmgen.assignRegister(RegisterOrPair.FAC1, assign.target)
+                return true
+            }
+            "/" -> {
+                assignFloatOperandsToFACandARG(expr.right, expr.left)
+                asmgen.out("  jsr  floats.FDIVT")
+                asmgen.assignRegister(RegisterOrPair.FAC1, assign.target)
+                return true
+            }
+            "==" -> {
+                setupFloatComparisonFAC1vsVarAY(expr)
+                asmgen.out("  jsr  floats.var_fac1_equal_f")
+                asmgen.assignRegister(RegisterOrPair.A, assign.target)
+                return true
+            }
+            "!=" -> {
+                setupFloatComparisonFAC1vsVarAY(expr)
+                asmgen.out("  jsr  floats.var_fac1_notequal_f")
+                asmgen.assignRegister(RegisterOrPair.A, assign.target)
+                return true
+            }
+            "<" -> {
+                setupFloatComparisonFAC1vsVarAY(expr)
+                asmgen.out("  jsr  floats.var_fac1_less_f")
+                asmgen.assignRegister(RegisterOrPair.A, assign.target)
+                return true
+            }
+            ">" -> {
+                setupFloatComparisonFAC1vsVarAY(expr)
+                asmgen.out("  jsr  floats.var_fac1_greater_f")
+                asmgen.assignRegister(RegisterOrPair.A, assign.target)
+                return true
+            }
+            "<=" -> {
+                setupFloatComparisonFAC1vsVarAY(expr)
+                asmgen.out("  jsr  floats.var_fac1_lesseq_f")
+                asmgen.assignRegister(RegisterOrPair.A, assign.target)
+                return true
+            }
+            ">=" -> {
+                setupFloatComparisonFAC1vsVarAY(expr)
+                asmgen.out("  jsr  floats.var_fac1_greatereq_f")
+                asmgen.assignRegister(RegisterOrPair.A, assign.target)
+                return true
+            }
+            else -> TODO("float expression operator ${expr.operator}")
+        }
+    }
+
+    private fun assignFloatOperandsToFACandARG(left: PtExpression, right: PtExpression) {
+        when(asmgen.options.compTarget.name) {
+            C64Target.NAME -> {
+                // c64 has a quirk: always make sure FAC2 is loaded last (done using CONUPK)  otherwise the result will be corrupt on C64
+                // this requires some more forced copying around of float values in certain cases
+                if (right.isSimple()) {
+                    asmgen.assignExpressionToRegister(left, RegisterOrPair.FAC1, true)
+                    asmgen.assignExpressionToRegister(right, RegisterOrPair.FAC2, true)
+                } else {
+                    asmgen.assignExpressionToRegister(right, RegisterOrPair.FAC1, true)
+                    asmgen.pushFAC1()
+                    asmgen.assignExpressionToRegister(left, RegisterOrPair.FAC1, true)
+                    asmgen.popFAC2()
+                }
+            }
+            Cx16Target.NAME -> {
+                asmgen.assignExpressionToRegister(left, RegisterOrPair.FAC1, true)
+                if (!right.isSimple()) asmgen.pushFAC1()
+                asmgen.assignExpressionToRegister(right, RegisterOrPair.FAC2, true)
+                if (!right.isSimple()) asmgen.popFAC1()
+            }
+            else -> TODO("don't know how to evaluate float expression for selected compilation target")
+        }
+    }
+
+    private fun setupFloatComparisonFAC1vsVarAY(expr: PtBinaryExpression) {
+        asmgen.assignExpressionToRegister(expr.left, RegisterOrPair.FAC1, true)
+        if(!expr.right.isSimple()) asmgen.pushFAC1()
+        asmgen.assignExpressionToVariable(expr.right, "floats.floats_temp_var", DataType.FLOAT)
+        if(!expr.right.isSimple()) asmgen.popFAC1()
+        asmgen.out("  lda  #<floats.floats_temp_var |  ldy  #>floats.floats_temp_var")
+    }
+}

codeGenCpu6502/src/prog8/codegen/cpu6502/assignment/AsmAssignment.kt

@@ -10,8 +10,7 @@ internal enum class TargetStorageKind {
     VARIABLE,
     ARRAY,
     MEMORY,
-    REGISTER,
-    STACK
+    REGISTER
 }
 
 internal enum class SourceStorageKind {
@@ -20,7 +19,6 @@ internal enum class SourceStorageKind {
     ARRAY,
     MEMORY,
     REGISTER,
-    STACK,              // value is already present on stack
     EXPRESSION,         // expression in ast-form, still to be evaluated
 }
 
@@ -120,7 +118,7 @@ internal class AsmAssignTarget(val kind: TargetStorageKind,
             TargetStorageKind.MEMORY -> {
                 left isSameAs memory!!
             }
-            TargetStorageKind.REGISTER, TargetStorageKind.STACK -> {
+            TargetStorageKind.REGISTER -> {
                 false
             }
         }
@@ -178,8 +176,8 @@ internal class AsmAssignSource(val kind: SourceStorageKind,
                     AsmAssignSource(SourceStorageKind.EXPRESSION, program, asmgen, value.type, expression = value)
                 }
                 is PtFunctionCall -> {
-                    val symbol = asmgen.symbolTable.lookup(value.name)
-                    val sub = symbol!!.astNode as IPtSubroutine
+                    val symbol = asmgen.symbolTable.lookup(value.name) ?: throw AssemblyError("lookup error ${value.name}")
+                    val sub = symbol.astNode as IPtSubroutine
                     val returnType = sub.returnsWhatWhere().firstOrNull { rr -> rr.first.registerOrPair != null || rr.first.statusflag!=null }?.second
                             ?: throw AssemblyError("can't translate zero return values in assignment")
 

codeGenCpu6502/test/Dummies.kt

@@ -45,6 +45,10 @@ internal class ErrorReporterForTests(private val throwExceptionAtReportIfErrors:
             warnings.add(text)
     }
 
+    override fun undefined(symbol: List<String>, position: Position) {
+        err("undefined symbol: ${symbol.joinToString(".")}", position)
+    }
+
     override fun noErrors(): Boolean  = errors.isEmpty()
 
     override fun report() {

codeGenCpu6502/test/TestCodegen.kt

@@ -1,7 +1,9 @@
 package prog8tests.codegencpu6502
 
 import io.kotest.assertions.throwables.shouldNotThrowAny
+import io.kotest.assertions.withClue
 import io.kotest.core.spec.style.FunSpec
+import io.kotest.matchers.ints.shouldBeGreaterThanOrEqual
 import io.kotest.matchers.shouldBe
 import prog8.code.SymbolTableMaker
 import prog8.code.ast.*
@@ -40,9 +42,9 @@ class TestCodegen: FunSpec({
 //        xx += cx16.r0
 //    }
 //}
-        val codegen = AsmGen6502()
+        val codegen = AsmGen6502(prefixSymbols = false)
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         sub.add(PtVariable("pi", DataType.UBYTE, ZeropageWish.DONTCARE, PtNumber(DataType.UBYTE, 0.0, Position.DUMMY), null, Position.DUMMY))
         sub.add(PtVariable("particleX", DataType.ARRAY_UB, ZeropageWish.DONTCARE, null, 3u, Position.DUMMY))
@@ -92,7 +94,7 @@ class TestCodegen: FunSpec({
         program.add(block)
 
         // define the "cx16.r0" virtual register
-        val cx16block = PtBlock("cx16", null, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val cx16block = PtBlock("cx16", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         cx16block.add(PtMemMapped("r0", DataType.UWORD, 100u, null, Position.DUMMY))
         program.add(cx16block)
 
@@ -104,12 +106,22 @@ class TestCodegen: FunSpec({
         Files.deleteIfExists(Path("${result.name}.asm"))
     }
 
-    test("64tass assembler available? - if this fails you need to install 64tass in the path") {
+    test("64tass assembler available? - if this fails you need to install 64tass version 1.58 or newer in the path") {
         val command = mutableListOf("64tass", "--version")
         shouldNotThrowAny {
-            val proc = ProcessBuilder(command).inheritIO().start()
+            val proc = ProcessBuilder(command).start()
+            val output = String(proc.inputStream.readBytes())
             val result = proc.waitFor()
             result.shouldBe(0)
+            val (_, version) = output.split('V')
+            val (major, minor, _) = version.split('.')
+            val majorNum = major.toInt()
+            val minorNum = minor.toInt()
+            withClue("64tass version should be 1.58 or newer") {
+                majorNum shouldBeGreaterThanOrEqual 1
+                if (majorNum == 1)
+                    minorNum shouldBeGreaterThanOrEqual 58
+            }
         }
     }
 })

codeGenExperimental/build.gradle

@@ -29,7 +29,7 @@ dependencies {
     implementation project(':codeGenIntermediate')
     implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8"
     // implementation "org.jetbrains.kotlin:kotlin-reflect"
-    implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.17"
+    implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.18"
 
 }
 

codeGenExperimental/src/prog8/codegen/experimental/ExperiCodeGen.kt

@@ -15,7 +15,7 @@ class ExperiCodeGen: ICodeGeneratorBackend {
         symbolTable: SymbolTable,
         options: CompilationOptions,
         errors: IErrorReporter
-    ): IAssemblyProgram? {
+    ): IAssemblyProgram {
 
         // you could write a code generator directly on the PtProgram AST,
         // but you can also use the Intermediate Representation to build a codegen on:
@@ -26,6 +26,15 @@ class ExperiCodeGen: ICodeGeneratorBackend {
         IRFileWriter(irProgram, null).write()
 
         println("** experimental codegen stub: no assembly generated **")
-        return null
+        return EmptyProgram
     }
-}
+}
+
+private object EmptyProgram : IAssemblyProgram {
+    override val name = "<Empty Program>"
+    override fun assemble(options: CompilationOptions, errors: IErrorReporter): Boolean {
+        println("** nothing assembled **")
+        return true
+    }
+
+}

codeGenIntermediate/build.gradle

@@ -28,7 +28,7 @@ dependencies {
     implementation project(':intermediate')
     implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8"
     // implementation "org.jetbrains.kotlin:kotlin-reflect"
-    implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.17"
+    implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.18"
 
     testImplementation 'io.kotest:kotest-runner-junit5-jvm:5.6.2'
 }

codeGenIntermediate/src/prog8/codegen/intermediate/AssignmentGen.kt

@@ -1,6 +1,5 @@
 package prog8.codegen.intermediate
 
-import prog8.code.StStaticVariable
 import prog8.code.ast.*
 import prog8.code.core.AssemblyError
 import prog8.code.core.DataType
@@ -14,7 +13,9 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
         if(assignment.target.children.single() is PtMachineRegister)
             throw AssemblyError("assigning to a register should be done by just evaluating the expression into resultregister")
 
-        return translateRegularAssign(assignment)
+        val chunks = translateRegularAssign(assignment)
+        chunks.filterIsInstance<IRCodeChunk>().firstOrNull()?.appendSrcPosition(assignment.position)
+        return chunks
     }
 
     internal fun translate(augAssign: PtAugmentedAssign): IRCodeChunks {
@@ -25,7 +26,7 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
         val memory = augAssign.target.memory
         val array = augAssign.target.array
 
-        return if(ident!=null) {
+        val chunks = if(ident!=null) {
             assignVarAugmented(ident.name, augAssign)
         } else if(memory != null) {
             if(memory.address is PtNumber)
@@ -40,6 +41,8 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
         } else {
             fallbackAssign(augAssign)
         }
+        chunks.filterIsInstance<IRCodeChunk>().firstOrNull()?.appendSrcPosition(augAssign.position)
+        return chunks
     }
 
     private fun assignMemoryAugmented(
@@ -49,15 +52,15 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
         val value = assignment.value
         val vmDt = irType(value.type)
         return when(assignment.operator) {
-            "+" -> expressionEval.operatorPlusInplace(address, null, vmDt, value)
-            "-" -> expressionEval.operatorMinusInplace(address, null, vmDt, value)
-            "*" -> expressionEval.operatorMultiplyInplace(address, null, vmDt, value)
-            "/" -> expressionEval.operatorDivideInplace(address, null, vmDt, value.type in SignedDatatypes, value)
-            "|" -> expressionEval.operatorOrInplace(address, null, vmDt, value)
-            "&" -> expressionEval.operatorAndInplace(address, null, vmDt, value)
-            "^" -> expressionEval.operatorXorInplace(address, null, vmDt, value)
-            "<<" -> expressionEval.operatorShiftLeftInplace(address, null, vmDt, value)
-            ">>" -> expressionEval.operatorShiftRightInplace(address, null, vmDt, value.type in SignedDatatypes, value)
+            "+=" -> expressionEval.operatorPlusInplace(address, null, vmDt, value)
+            "-=" -> expressionEval.operatorMinusInplace(address, null, vmDt, value)
+            "*=" -> expressionEval.operatorMultiplyInplace(address, null, vmDt, value)
+            "/=" -> expressionEval.operatorDivideInplace(address, null, vmDt, value.type in SignedDatatypes, value)
+            "|=" -> expressionEval.operatorOrInplace(address, null, vmDt, value)
+            "&=" -> expressionEval.operatorAndInplace(address, null, vmDt, value)
+            "^=" -> expressionEval.operatorXorInplace(address, null, vmDt, value)
+            "<<=" -> expressionEval.operatorShiftLeftInplace(address, null, vmDt, value)
+            ">>=" -> expressionEval.operatorShiftRightInplace(address, null, vmDt, value.type in SignedDatatypes, value)
             "%=" -> expressionEval.operatorModuloInplace(address, null, vmDt, value)
             "==" -> expressionEval.operatorEqualsInplace(address, null, vmDt, value)
             "!=" -> expressionEval.operatorNotEqualsInplace(address, null, vmDt, value)
@@ -97,35 +100,16 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
     }
 
     private fun fallbackAssign(origAssign: PtAugmentedAssign): IRCodeChunks {
-        if (codeGen.options.slowCodegenWarnings)
-            codeGen.errors.warn("indirect code for in-place assignment", origAssign.position)
         val value: PtExpression
         if(origAssign.operator in PrefixOperators) {
             value = PtPrefix(origAssign.operator, origAssign.value.type, origAssign.value.position)
             value.add(origAssign.value)
         } else {
             require(origAssign.operator.endsWith('='))
-            if(codeGen.options.useNewExprCode) {
-                // X += Y  ->   temp = X,  temp += Y,  X = temp
-                val tempvar = codeGen.getReusableTempvar(origAssign.definingSub()!!, origAssign.target.type)
-                val assign = PtAssignment(origAssign.position)
-                val target = PtAssignTarget(origAssign.position)
-                target.add(tempvar)
-                assign.add(target)
-                assign.add(origAssign.target.children.single())
-                val augAssign = PtAugmentedAssign(origAssign.operator, origAssign.position)
-                augAssign.add(target)
-                augAssign.add(origAssign.value)
-                val assignBack = PtAssignment(origAssign.position)
-                assignBack.add(origAssign.target)
-                assignBack.add(tempvar)
-                return translateRegularAssign(assign) + translate(augAssign) + translateRegularAssign(assignBack)
-            } else {
-                value = PtBinaryExpression(origAssign.operator.dropLast(1), origAssign.value.type, origAssign.value.position)
-                val left: PtExpression = origAssign.target.children.single() as PtExpression
-                value.add(left)
-                value.add(origAssign.value)
-            }
+            value = PtBinaryExpression(origAssign.operator.dropLast(1), origAssign.value.type, origAssign.value.position)
+            val left: PtExpression = origAssign.target.children.single() as PtExpression
+            value.add(left)
+            value.add(origAssign.value)
         }
         val normalAssign = PtAssignment(origAssign.position)
         normalAssign.add(origAssign.target)
@@ -185,15 +169,18 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
                 } else false
                 if (assignment.value is PtMachineRegister) {
                     valueRegister = (assignment.value as PtMachineRegister).register
-                    if(extendByteToWord)
-                        addInstr(result, IRInstruction(Opcode.EXT, IRDataType.BYTE, reg1=valueRegister), null)
+                    if(extendByteToWord) {
+                        valueRegister = codeGen.registers.nextFree()
+                        addInstr(result, IRInstruction(Opcode.EXT, IRDataType.BYTE, reg1=valueRegister, reg2=(assignment.value as PtMachineRegister).register), null)
+                    }
                 } else {
                     val tr = expressionEval.translateExpression(assignment.value)
                     valueRegister = tr.resultReg
                     addToResult(result, tr, valueRegister, -1)
                     if(extendByteToWord) {
+                        valueRegister = codeGen.registers.nextFree()
                         val opcode = if(assignment.value.type in SignedDatatypes) Opcode.EXTS else Opcode.EXT
-                        addInstr(result, IRInstruction(opcode, IRDataType.BYTE, reg1 = valueRegister), null)
+                        addInstr(result, IRInstruction(opcode, IRDataType.BYTE, reg1=valueRegister, reg2=tr.resultReg), null)
                     }
                 }
             }
@@ -215,8 +202,7 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
             val variable = targetArray.variable.name
             val itemsize = codeGen.program.memsizer.memorySize(targetArray.type)
 
-            if(targetArray.variable.type==DataType.UWORD) {
-                // indexing a pointer var instead of a real array or string
+            if(targetArray.usesPointerVariable) {
                 if(itemsize!=1)
                     throw AssemblyError("non-array var indexing requires bytes dt")
                 if(targetArray.index.type!=DataType.UBYTE)
@@ -236,8 +222,7 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
             }
 
             val fixedIndex = constIntValue(targetArray.index)
-            val iterable = codeGen.symbolTable.flat.getValue(targetArray.variable.name) as StStaticVariable
-            val arrayLength = iterable.length!!
+            val arrayLength = codeGen.symbolTable.getLength(targetArray.variable.name)
             if(zero) {
                 if(fixedIndex!=null) {
                     val chunk = IRCodeChunk(null, null).also {
@@ -347,19 +332,12 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
             return Pair(result, tr.resultReg)
         }
 
-        if(codeGen.options.useNewExprCode) {
-            val tr = expressionEval.translateExpression(array.index)
-            result += tr.chunks
-            addInstr(result, IRInstruction(Opcode.MUL, tr.dt, reg1=tr.resultReg, immediate = itemsize), null)
-            return Pair(result, tr.resultReg)
-        } else {
-            val mult: PtExpression
-            mult = PtBinaryExpression("*", DataType.UBYTE, array.position)
-            mult.children += array.index
-            mult.children += PtNumber(DataType.UBYTE, itemsize.toDouble(), array.position)
-            val tr = expressionEval.translateExpression(mult)
-            addToResult(result, tr, tr.resultReg, -1)
-            return Pair(result, tr.resultReg)
-        }
+        val mult: PtExpression
+        mult = PtBinaryExpression("*", DataType.UBYTE, array.position)
+        mult.children += array.index
+        mult.children += PtNumber(DataType.UBYTE, itemsize.toDouble(), array.position)
+        val tr = expressionEval.translateExpression(mult)
+        addToResult(result, tr, tr.resultReg, -1)
+        return Pair(result, tr.resultReg)
     }
 }

codeGenIntermediate/src/prog8/codegen/intermediate/BuiltinFuncGen.kt

@@ -1,6 +1,5 @@
 package prog8.codegen.intermediate
 
-import prog8.code.StStaticVariable
 import prog8.code.ast.*
 import prog8.code.core.AssemblyError
 import prog8.code.core.DataType
@@ -25,11 +24,8 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
             "popw" -> funcPopw(call)
             "push" -> funcPush(call)
             "pushw" -> funcPushw(call)
-            "rsave",
-            "rsavex",
-            "rrestore",
-            "rrestorex" -> ExpressionCodeResult.EMPTY  // vm doesn't have registers to save/restore
-            "callfar" -> throw AssemblyError("callfar() is for cx16 target only")
+            "rsave", "rrestore" -> ExpressionCodeResult.EMPTY  // vm doesn't have registers to save/restore
+            "callfar" -> funcCallfar(call)
             "msb" -> funcMsb(call)
             "lsb" -> funcLsb(call)
             "memory" -> funcMemory(call)
@@ -49,10 +45,40 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
             "rol2" -> funcRolRor(Opcode.ROL, call)
             "ror2" -> funcRolRor(Opcode.ROR, call)
             "prog8_lib_stringcompare" -> funcStringCompare(call)
+            "prog8_lib_square_byte" -> funcSquare(call, IRDataType.BYTE)
+            "prog8_lib_square_word" -> funcSquare(call, IRDataType.WORD)
             else -> throw AssemblyError("missing builtinfunc for ${call.name}")
         }
     }
 
+    private fun funcSquare(call: PtBuiltinFunctionCall, resultType: IRDataType): ExpressionCodeResult {
+        val result = mutableListOf<IRCodeChunkBase>()
+        val valueTr = exprGen.translateExpression(call.args[0])
+        addToResult(result, valueTr, valueTr.resultReg, valueTr.resultFpReg)
+        return if(resultType==IRDataType.FLOAT) {
+            val resultFpReg = codeGen.registers.nextFreeFloat()
+            addInstr(result, IRInstruction(Opcode.SQUARE, resultType, fpReg1 = resultFpReg, fpReg2 = valueTr.resultFpReg), null)
+            ExpressionCodeResult(result, resultType, -1, resultFpReg)
+        }
+        else {
+            val resultReg = codeGen.registers.nextFree()
+            addInstr(result, IRInstruction(Opcode.SQUARE, resultType, reg1 = resultReg, reg2 = valueTr.resultReg), null)
+            ExpressionCodeResult(result, resultType, resultReg, -1)
+        }
+    }
+
+    private fun funcCallfar(call: PtBuiltinFunctionCall): ExpressionCodeResult {
+        val result = mutableListOf<IRCodeChunkBase>()
+        val bankTr = exprGen.translateExpression(call.args[0])
+        val addressTr = exprGen.translateExpression(call.args[1])
+        val argumentwordTr = exprGen.translateExpression(call.args[2])
+        addToResult(result, bankTr, bankTr.resultReg, -1)
+        addToResult(result, addressTr, addressTr.resultReg, -1)
+        addToResult(result, argumentwordTr, argumentwordTr.resultReg, -1)
+        result += codeGen.makeSyscall(IMSyscall.CALLFAR, listOf(IRDataType.BYTE to bankTr.resultReg, IRDataType.WORD to addressTr.resultReg, IRDataType.WORD to argumentwordTr.resultReg), IRDataType.WORD to argumentwordTr.resultReg)
+        return ExpressionCodeResult(result, IRDataType.WORD, argumentwordTr.resultReg, -1)
+    }
+
     private fun funcDivmod(call: PtBuiltinFunctionCall, type: IRDataType): ExpressionCodeResult {
         val result = mutableListOf<IRCodeChunkBase>()
         val number = call.args[0]
@@ -107,9 +133,9 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
 
     private fun funcAny(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         val arrayName = call.args[0] as PtIdentifier
-        val array = codeGen.symbolTable.flat.getValue(arrayName.name) as StStaticVariable
+        val arrayLength = codeGen.symbolTable.getLength(arrayName.name)
         val syscall =
-            when (array.dt) {
+            when (arrayName.type) {
                 DataType.ARRAY_UB,
                 DataType.ARRAY_B -> IMSyscall.ANY_BYTE
                 DataType.ARRAY_UW,
@@ -121,16 +147,16 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
         val tr = exprGen.translateExpression(call.args[0])
         addToResult(result, tr, tr.resultReg, -1)
         val lengthReg = codeGen.registers.nextFree()
-        addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = lengthReg, immediate = array.length), null)
+        addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = lengthReg, immediate = arrayLength), null)
         result += codeGen.makeSyscall(syscall, listOf(IRDataType.WORD to tr.resultReg, IRDataType.BYTE to lengthReg), IRDataType.BYTE to tr.resultReg)
         return ExpressionCodeResult(result, IRDataType.BYTE, tr.resultReg, -1)
     }
 
     private fun funcAll(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         val arrayName = call.args[0] as PtIdentifier
-        val array = codeGen.symbolTable.flat.getValue(arrayName.name) as StStaticVariable
+        val arrayLength = codeGen.symbolTable.getLength(arrayName.name)
         val syscall =
-            when(array.dt) {
+            when(arrayName.type) {
                 DataType.ARRAY_UB,
                 DataType.ARRAY_B -> IMSyscall.ALL_BYTE
                 DataType.ARRAY_UW,
@@ -142,7 +168,7 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
         val tr = exprGen.translateExpression(call.args[0])
         addToResult(result, tr, tr.resultReg, -1)
         val lengthReg = codeGen.registers.nextFree()
-        addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = lengthReg, immediate = array.length), null)
+        addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = lengthReg, immediate = arrayLength), null)
         result += codeGen.makeSyscall(syscall, listOf(IRDataType.WORD to tr.resultReg, IRDataType.BYTE to lengthReg), IRDataType.BYTE to tr.resultReg)
         return ExpressionCodeResult(result, IRDataType.BYTE, tr.resultReg, -1)
     }
@@ -161,10 +187,8 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
                 val compareReg = codeGen.registers.nextFree()
                 result += IRCodeChunk(null, null).also {
                     it += IRInstruction(Opcode.LOADR, IRDataType.BYTE, reg1=compareReg, reg2=tr.resultReg)
-                    it += IRInstruction(Opcode.AND, IRDataType.BYTE, reg1=compareReg, immediate = 0x80)
-                    it += IRInstruction(Opcode.BEQ, IRDataType.BYTE, reg1=compareReg, immediate = 0, labelSymbol = notNegativeLabel)
+                    it += IRInstruction(Opcode.BSTPOS, labelSymbol = notNegativeLabel)
                     it += IRInstruction(Opcode.NEG, IRDataType.BYTE, reg1=tr.resultReg)
-                    it += IRInstruction(Opcode.EXT, IRDataType.BYTE, reg1=tr.resultReg)
                 }
                 result += IRCodeChunk(notNegativeLabel, null)
                 return ExpressionCodeResult(result, IRDataType.BYTE, tr.resultReg, -1)
@@ -174,8 +198,7 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
                 val compareReg = codeGen.registers.nextFree()
                 result += IRCodeChunk(null, null).also {
                     it += IRInstruction(Opcode.LOADR, IRDataType.WORD, reg1=compareReg, reg2=tr.resultReg)
-                    it += IRInstruction(Opcode.AND, IRDataType.WORD, reg1=compareReg, immediate = 0x8000)
-                    it += IRInstruction(Opcode.BEQ, IRDataType.WORD, reg1=compareReg, immediate = 0, labelSymbol = notNegativeLabel)
+                    it += IRInstruction(Opcode.BSTPOS, labelSymbol = notNegativeLabel)
                     it += IRInstruction(Opcode.NEG, IRDataType.WORD, reg1=tr.resultReg)
                 }
                 result += IRCodeChunk(notNegativeLabel, null)
@@ -275,9 +298,9 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
 
     private fun funcReverse(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         val arrayName = call.args[0] as PtIdentifier
-        val array = codeGen.symbolTable.flat.getValue(arrayName.name) as StStaticVariable
+        val arrayLength = codeGen.symbolTable.getLength(arrayName.name)
         val syscall =
-            when(array.dt) {
+            when(arrayName.type) {
                 DataType.ARRAY_UB, DataType.ARRAY_B, DataType.STR -> IMSyscall.REVERSE_BYTES
                 DataType.ARRAY_UW, DataType.ARRAY_W -> IMSyscall.REVERSE_WORDS
                 DataType.ARRAY_F -> IMSyscall.REVERSE_FLOATS
@@ -288,16 +311,16 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
         val tr = exprGen.translateExpression(call.args[0])
         addToResult(result, tr, tr.resultReg, -1)
         val lengthReg = codeGen.registers.nextFree()
-        addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = lengthReg, immediate = if(array.dt==DataType.STR) array.length!!-1 else array.length), null)
+        addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = lengthReg, immediate = if(arrayName.type==DataType.STR) arrayLength!!-1 else arrayLength), null)
         result += codeGen.makeSyscall(syscall, listOf(IRDataType.WORD to tr.resultReg, IRDataType.BYTE to lengthReg), null)
         return ExpressionCodeResult(result, IRDataType.BYTE, -1, -1)
     }
 
     private fun funcSort(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         val arrayName = call.args[0] as PtIdentifier
-        val array = codeGen.symbolTable.flat.getValue(arrayName.name) as StStaticVariable
+        val arrayLength = codeGen.symbolTable.getLength(arrayName.name)
         val syscall =
-            when(array.dt) {
+            when(arrayName.type) {
                 DataType.ARRAY_UB -> IMSyscall.SORT_UBYTE
                 DataType.ARRAY_B -> IMSyscall.SORT_BYTE
                 DataType.ARRAY_UW -> IMSyscall.SORT_UWORD
@@ -311,7 +334,7 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
         val tr = exprGen.translateExpression(call.args[0])
         addToResult(result, tr, tr.resultReg, -1)
         val lengthReg = codeGen.registers.nextFree()
-        addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = lengthReg, immediate = if(array.dt==DataType.STR) array.length!!-1 else array.length), null)
+        addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = lengthReg, immediate = if(arrayName.type==DataType.STR) arrayLength!!-1 else arrayLength), null)
         result += codeGen.makeSyscall(syscall, listOf(IRDataType.WORD to tr.resultReg, IRDataType.BYTE to lengthReg), null)
         return ExpressionCodeResult(result, IRDataType.BYTE, -1, -1)
     }
@@ -322,10 +345,11 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
         addToResult(result, msbTr, msbTr.resultReg, -1)
         val lsbTr = exprGen.translateExpression(call.args[1])
         addToResult(result, lsbTr, lsbTr.resultReg, -1)
+        val resultReg = codeGen.registers.nextFree()
         result += IRCodeChunk(null, null).also {
-            it += IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1 = lsbTr.resultReg, reg2 = msbTr.resultReg)
+            it += IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=resultReg, reg2 = lsbTr.resultReg, reg3 = msbTr.resultReg)
         }
-        return ExpressionCodeResult(result, IRDataType.WORD, lsbTr.resultReg, -1)
+        return ExpressionCodeResult(result, IRDataType.WORD, resultReg, -1)
     }
 
     private fun funcClamp(call: PtBuiltinFunctionCall): ExpressionCodeResult {
@@ -525,6 +549,7 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
     private fun funcLsb(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         return exprGen.translateExpression(call.args.single())
         // note: if a word result is needed, the upper byte is cleared by the typecast that follows. No need to do it here.
+        // TODO to be more strict, maybe we *should* introduce a new result register that is of type .b?
     }
 
     private fun funcMsb(call: PtBuiltinFunctionCall): ExpressionCodeResult {

codeGenIntermediate/src/prog8/codegen/intermediate/ExpressionGen.kt

@@ -1,7 +1,6 @@
 package prog8.codegen.intermediate
 
 import prog8.code.StRomSub
-import prog8.code.StStaticVariable
 import prog8.code.StSub
 import prog8.code.ast.*
 import prog8.code.core.AssemblyError
@@ -45,9 +44,9 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 }
             }
             is PtIdentifier -> {
-                val vmDt = irType(expr.type)
                 val code = IRCodeChunk(null, null)
                 if (expr.type in PassByValueDatatypes) {
+                    val vmDt = irType(expr.type)
                     if(vmDt==IRDataType.FLOAT) {
                         val resultFpRegister = codeGen.registers.nextFreeFloat()
                         code += IRInstruction(Opcode.LOADM, vmDt, fpReg1 = resultFpRegister, labelSymbol = expr.name)
@@ -60,6 +59,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                     }
                 } else {
                     // for strings and arrays etc., load the *address* of the value instead
+                    val vmDt = if(expr.type==DataType.UNDEFINED) IRDataType.WORD else irType(expr.type)
                     val resultRegister = codeGen.registers.nextFree()
                     code += IRInstruction(Opcode.LOAD, vmDt, reg1 = resultRegister, labelSymbol = expr.name)
                     ExpressionCodeResult(code, vmDt, resultRegister, -1)
@@ -105,8 +105,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
 
     private fun translate(check: PtContainmentCheck): ExpressionCodeResult {
         val result = mutableListOf<IRCodeChunkBase>()
-        val iterable = codeGen.symbolTable.flat.getValue(check.iterable.name) as StStaticVariable
-        when(iterable.dt) {
+        when(check.iterable.type) {
             DataType.STR -> {
                 val elementTr = translateExpression(check.element)
                 addToResult(result, elementTr, elementTr.resultReg, -1)
@@ -121,7 +120,8 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 val iterableTr = translateExpression(check.iterable)
                 addToResult(result, iterableTr, iterableTr.resultReg, -1)
                 val lengthReg = codeGen.registers.nextFree()
-                addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=lengthReg, immediate = iterable.length!!), null)
+                val iterableLength = codeGen.symbolTable.getLength(check.iterable.name)
+                addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=lengthReg, immediate = iterableLength!!), null)
                 result += codeGen.makeSyscall(IMSyscall.BYTEARRAY_CONTAINS, listOf(IRDataType.BYTE to elementTr.resultReg, IRDataType.WORD to iterableTr.resultReg, IRDataType.BYTE to lengthReg), IRDataType.BYTE to elementTr.resultReg)
                 return ExpressionCodeResult(result, IRDataType.BYTE, elementTr.resultReg, -1)
             }
@@ -131,12 +131,13 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 val iterableTr = translateExpression(check.iterable)
                 addToResult(result, iterableTr, iterableTr.resultReg, -1)
                 val lengthReg = codeGen.registers.nextFree()
-                addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=lengthReg, immediate = iterable.length!!), null)
+                val iterableLength = codeGen.symbolTable.getLength(check.iterable.name)
+                addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=lengthReg, immediate = iterableLength!!), null)
                 result += codeGen.makeSyscall(IMSyscall.WORDARRAY_CONTAINS, listOf(IRDataType.WORD to elementTr.resultReg, IRDataType.WORD to iterableTr.resultReg, IRDataType.BYTE to lengthReg), IRDataType.BYTE to elementTr.resultReg)
                 return ExpressionCodeResult(result, IRDataType.BYTE, elementTr.resultReg, -1)
             }
             DataType.ARRAY_F -> throw AssemblyError("containment check in float-array not supported")
-            else -> throw AssemblyError("weird iterable dt ${iterable.dt} for ${check.iterable.name}")
+            else -> throw AssemblyError("weird iterable dt ${check.iterable.type} for ${check.iterable.name}")
         }
     }
 
@@ -146,8 +147,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
         val result = mutableListOf<IRCodeChunkBase>()
         val arrayVarSymbol = arrayIx.variable.name
 
-        if(arrayIx.variable.type==DataType.UWORD) {
-            // indexing a pointer var instead of a real array or string
+        if(arrayIx.usesPointerVariable) {
             if(eltSize!=1)
                 throw AssemblyError("non-array var indexing requires bytes dt")
             if(arrayIx.index.type!=DataType.UBYTE)
@@ -163,16 +163,16 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
 
         if(arrayIx.splitWords) {
             require(vmDt==IRDataType.WORD)
-            val iterable = codeGen.symbolTable.flat.getValue(arrayIx.variable.name) as StStaticVariable
-            val arrayLength = iterable.length!!
+            val arrayLength = codeGen.symbolTable.getLength(arrayIx.variable.name)
             resultRegister = codeGen.registers.nextFree()
+            val finalResultReg = codeGen.registers.nextFree()
             if(arrayIx.index is PtNumber) {
                 val memOffset = (arrayIx.index as PtNumber).number.toInt()
                 result += IRCodeChunk(null, null).also {
                     val tmpRegMsb = codeGen.registers.nextFree()
                     it += IRInstruction(Opcode.LOADM, IRDataType.BYTE, reg1=resultRegister, immediate = arrayLength, labelSymbol= "${arrayVarSymbol}_lsb+$memOffset")
                     it += IRInstruction(Opcode.LOADM, IRDataType.BYTE, reg1=tmpRegMsb, immediate = arrayLength, labelSymbol= "${arrayVarSymbol}_msb+$memOffset")
-                    it += IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=resultRegister, reg2=tmpRegMsb)
+                    it += IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=finalResultReg, reg2=resultRegister, reg3=tmpRegMsb)
                 }
             } else {
                 val tr = translateExpression(arrayIx.index)
@@ -181,10 +181,10 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                     val tmpRegMsb = codeGen.registers.nextFree()
                     it += IRInstruction(Opcode.LOADX, IRDataType.BYTE, reg1=resultRegister, reg2 = tr.resultReg, immediate = arrayLength, labelSymbol= "${arrayVarSymbol}_lsb")
                     it += IRInstruction(Opcode.LOADX, IRDataType.BYTE, reg1=tmpRegMsb, reg2 = tr.resultReg, immediate = arrayLength, labelSymbol= "${arrayVarSymbol}_msb")
-                    it += IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=resultRegister, reg2=tmpRegMsb)
+                    it += IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=finalResultReg, reg2=resultRegister, reg3=tmpRegMsb)
                 }
             }
-            return ExpressionCodeResult(result, vmDt, resultRegister, -1)
+            return ExpressionCodeResult(result, vmDt, finalResultReg, -1)
         }
 
         var resultFpRegister = -1
@@ -274,13 +274,13 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 when(cast.value.type) {
                     DataType.BYTE -> {
                         // byte -> uword:   sign extend
-                        actualResultReg2 = tr.resultReg
-                        addInstr(result, IRInstruction(Opcode.EXTS, type = IRDataType.BYTE, reg1 = actualResultReg2), null)
+                        actualResultReg2 = codeGen.registers.nextFree()
+                        addInstr(result, IRInstruction(Opcode.EXTS, type = IRDataType.BYTE, reg1 = actualResultReg2, reg2 = tr.resultReg), null)
                     }
                     DataType.UBYTE -> {
                         // ubyte -> uword:   sign extend
-                        actualResultReg2 = tr.resultReg
-                        addInstr(result, IRInstruction(Opcode.EXT, type = IRDataType.BYTE, reg1 = actualResultReg2), null)
+                        actualResultReg2 = codeGen.registers.nextFree()
+                        addInstr(result, IRInstruction(Opcode.EXT, type = IRDataType.BYTE, reg1 = actualResultReg2, reg2 = tr.resultReg), null)
                     }
                     DataType.WORD -> {
                         actualResultReg2 = tr.resultReg
@@ -296,13 +296,13 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 when(cast.value.type) {
                     DataType.BYTE -> {
                         // byte -> word:   sign extend
-                        actualResultReg2 = tr.resultReg
-                        addInstr(result, IRInstruction(Opcode.EXTS, type = IRDataType.BYTE, reg1 = actualResultReg2), null)
+                        actualResultReg2 = codeGen.registers.nextFree()
+                        addInstr(result, IRInstruction(Opcode.EXTS, type = IRDataType.BYTE, reg1 = actualResultReg2, reg2=tr.resultReg), null)
                     }
                     DataType.UBYTE -> {
                         // byte -> word:   sign extend
-                        actualResultReg2 = tr.resultReg
-                        addInstr(result, IRInstruction(Opcode.EXT, type = IRDataType.BYTE, reg1 = actualResultReg2), null)
+                        actualResultReg2 = codeGen.registers.nextFree()
+                        addInstr(result, IRInstruction(Opcode.EXT, type = IRDataType.BYTE, reg1 = actualResultReg2, reg2=tr.resultReg), null)
                     }
                     DataType.UWORD -> {
                         actualResultReg2 = tr.resultReg
@@ -339,7 +339,6 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
     }
 
     private fun translate(binExpr: PtBinaryExpression): ExpressionCodeResult {
-        require(!codeGen.options.useNewExprCode)
         val vmDt = irType(binExpr.left.type)
         val signed = binExpr.left.type in SignedDatatypes
         return when(binExpr.operator) {
@@ -455,6 +454,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
         greaterEquals: Boolean
     ): ExpressionCodeResult {
         val result = mutableListOf<IRCodeChunkBase>()
+        val cmpResultReg = codeGen.registers.nextFree()
         if(vmDt==IRDataType.FLOAT) {
             val leftTr = translateExpression(binExpr.left)
             addToResult(result, leftTr, -1, leftTr.resultFpReg)
@@ -469,10 +469,10 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
             } else {
                 if (greaterEquals) Opcode.SGE else Opcode.SGT
             }
-            addInstr(result, IRInstruction(ins, IRDataType.BYTE, reg1 = resultRegister, reg2 = zeroRegister), null)
-            return ExpressionCodeResult(result, IRDataType.BYTE, resultRegister, -1)
+            addInstr(result, IRInstruction(ins, IRDataType.BYTE, reg1=cmpResultReg, reg2 = resultRegister, reg3 = zeroRegister), null)
+            return ExpressionCodeResult(result, IRDataType.BYTE, cmpResultReg, -1)
         } else {
-            if(binExpr.left.type==DataType.STR && binExpr.right.type==DataType.STR) {
+            if(binExpr.left.type==DataType.STR || binExpr.right.type==DataType.STR) {
                 throw AssemblyError("str compares should have been replaced with builtin function call to do the compare")
             } else {
                 val leftTr = translateExpression(binExpr.left)
@@ -484,8 +484,8 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 } else {
                     if (greaterEquals) Opcode.SGE else Opcode.SGT
                 }
-                addInstr(result, IRInstruction(ins, vmDt, reg1 = leftTr.resultReg, reg2 = rightTr.resultReg), null)
-                return ExpressionCodeResult(result, IRDataType.BYTE, leftTr.resultReg, -1)
+                addInstr(result, IRInstruction(ins, vmDt, reg1=cmpResultReg, reg2 = leftTr.resultReg, reg3 = rightTr.resultReg), null)
+                return ExpressionCodeResult(result, IRDataType.BYTE, cmpResultReg, -1)
             }
         }
     }
@@ -497,6 +497,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
         lessEquals: Boolean
     ): ExpressionCodeResult {
         val result = mutableListOf<IRCodeChunkBase>()
+        val cmpResultRegister = codeGen.registers.nextFree()
         if(vmDt==IRDataType.FLOAT) {
             val leftTr = translateExpression(binExpr.left)
             addToResult(result, leftTr, -1, leftTr.resultFpReg)
@@ -511,10 +512,10 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
             } else {
                 if (lessEquals) Opcode.SLE else Opcode.SLT
             }
-            addInstr(result, IRInstruction(ins, IRDataType.BYTE, reg1 = resultRegister, reg2 = zeroRegister), null)
-            return ExpressionCodeResult(result, IRDataType.BYTE, resultRegister, -1)
+            addInstr(result, IRInstruction(ins, IRDataType.BYTE, reg1=cmpResultRegister, reg2 = resultRegister, reg3 = zeroRegister), null)
+            return ExpressionCodeResult(result, IRDataType.BYTE, cmpResultRegister, -1)
         } else {
-            if(binExpr.left.type==DataType.STR && binExpr.right.type==DataType.STR) {
+            if(binExpr.left.type==DataType.STR || binExpr.right.type==DataType.STR) {
                 throw AssemblyError("str compares should have been replaced with builtin function call to do the compare")
             } else {
                 val leftTr = translateExpression(binExpr.left)
@@ -526,8 +527,8 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 } else {
                     if (lessEquals) Opcode.SLE else Opcode.SLT
                 }
-                addInstr(result, IRInstruction(ins, vmDt, reg1 = leftTr.resultReg, reg2 = rightTr.resultReg), null)
-                return ExpressionCodeResult(result, IRDataType.BYTE, leftTr.resultReg, -1)
+                addInstr(result, IRInstruction(ins, vmDt, reg1=cmpResultRegister, reg2 = leftTr.resultReg, reg3 = rightTr.resultReg), null)
+                return ExpressionCodeResult(result, IRDataType.BYTE, cmpResultRegister, -1)
             }
         }
     }
@@ -553,7 +554,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
             result += IRCodeChunk(label, null)
             return ExpressionCodeResult(result, IRDataType.BYTE, resultRegister, -1)
         } else {
-            if(binExpr.left.type==DataType.STR && binExpr.right.type==DataType.STR) {
+            if(binExpr.left.type==DataType.STR || binExpr.right.type==DataType.STR) {
                 throw AssemblyError("str compares should have been replaced with builtin function call to do the compare")
             } else {
                 return if(constValue(binExpr.right)==0.0) {
@@ -569,8 +570,9 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                     val rightTr = translateExpression(binExpr.right)
                     addToResult(result, rightTr, rightTr.resultReg, -1)
                     val opcode = if (notEquals) Opcode.SNE else Opcode.SEQ
-                    addInstr(result, IRInstruction(opcode, vmDt, reg1 = leftTr.resultReg, reg2 = rightTr.resultReg), null)
-                    ExpressionCodeResult(result, IRDataType.BYTE, leftTr.resultReg, -1)
+                    val resultReg = codeGen.registers.nextFree()
+                    addInstr(result, IRInstruction(opcode, vmDt, reg1 = resultReg, reg2 = leftTr.resultReg, reg3 = rightTr.resultReg), null)
+                    ExpressionCodeResult(result, IRDataType.BYTE, resultReg, -1)
                 }
             }
         }
@@ -1240,6 +1242,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
     ): MutableList<IRCodeChunkBase> {
         val result = mutableListOf<IRCodeChunkBase>()
         val valueReg = codeGen.registers.nextFree()
+        val cmpResultReg = codeGen.registers.nextFree()
         if(operand is PtNumber) {
             val numberReg = codeGen.registers.nextFree()
             if (knownAddress != null) {
@@ -1247,16 +1250,16 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 result += IRCodeChunk(null, null).also {
                     it += IRInstruction(Opcode.LOADM, vmDt, reg1 = valueReg, address = knownAddress)
                     it += IRInstruction(Opcode.LOAD, vmDt, reg1=numberReg, immediate = operand.number.toInt())
-                    it += IRInstruction(compareAndSetOpcode, vmDt, reg1 = valueReg, reg2 = numberReg)
-                    it += IRInstruction(Opcode.STOREM, vmDt, reg1 = valueReg, address = knownAddress)
+                    it += IRInstruction(compareAndSetOpcode, vmDt, reg1 = cmpResultReg, reg2 = valueReg, reg3 = numberReg)
+                    it += IRInstruction(Opcode.STOREM, vmDt, reg1 = cmpResultReg, address = knownAddress)
                 }
             } else {
                 // in-place modify a symbol (variable)
                 result += IRCodeChunk(null, null).also {
                     it += IRInstruction(Opcode.LOADM, vmDt, reg1 = valueReg, labelSymbol = symbol)
                     it += IRInstruction(Opcode.LOAD, vmDt, reg1=numberReg, immediate = operand.number.toInt())
-                    it += IRInstruction(compareAndSetOpcode, vmDt, reg1 = valueReg, reg2 = numberReg)
-                    it += IRInstruction(Opcode.STOREM, vmDt, reg1 = valueReg, labelSymbol = symbol)
+                    it += IRInstruction(compareAndSetOpcode, vmDt, reg1=cmpResultReg, reg2 = valueReg, reg3 = numberReg)
+                    it += IRInstruction(Opcode.STOREM, vmDt, reg1 = cmpResultReg, labelSymbol = symbol)
                 }
             }
         } else {
@@ -1266,15 +1269,15 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 // in-place modify a memory location
                 result += IRCodeChunk(null, null).also {
                     it += IRInstruction(Opcode.LOADM, vmDt, reg1 = valueReg, address = knownAddress)
-                    it += IRInstruction(compareAndSetOpcode, vmDt, reg1 = valueReg, reg2 = tr.resultReg)
-                    it += IRInstruction(Opcode.STOREM, vmDt, reg1 = valueReg, address = knownAddress)
+                    it += IRInstruction(compareAndSetOpcode, vmDt, reg1=cmpResultReg, reg2 = valueReg, reg3 = tr.resultReg)
+                    it += IRInstruction(Opcode.STOREM, vmDt, reg1 = cmpResultReg, address = knownAddress)
                 }
             } else {
                 // in-place modify a symbol (variable)
                 result += IRCodeChunk(null, null).also {
                     it += IRInstruction(Opcode.LOADM, vmDt, reg1 = valueReg, labelSymbol = symbol)
-                    it += IRInstruction(compareAndSetOpcode, vmDt, reg1 = valueReg, reg2 = tr.resultReg)
-                    it += IRInstruction(Opcode.STOREM, vmDt, reg1 = valueReg, labelSymbol = symbol)
+                    it += IRInstruction(compareAndSetOpcode, vmDt, reg1=cmpResultReg, reg2 = valueReg, reg3 = tr.resultReg)
+                    it += IRInstruction(Opcode.STOREM, vmDt, reg1 = cmpResultReg, labelSymbol = symbol)
                 }
             }
         }
@@ -1293,6 +1296,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
         val zeroReg = codeGen.registers.nextFree()
         if(operand is PtNumber) {
             val numberReg = codeGen.registers.nextFreeFloat()
+            val cmpResultReg = codeGen.registers.nextFree()
             if (knownAddress != null) {
                 // in-place modify a memory location
                 result += IRCodeChunk(null, null).also {
@@ -1300,8 +1304,8 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                     it += IRInstruction(Opcode.LOAD, IRDataType.FLOAT, fpReg1 = numberReg, immediateFp = operand.number.toFloat())
                     it += IRInstruction(Opcode.FCOMP, IRDataType.FLOAT, reg1=cmpReg, fpReg1 = valueReg, fpReg2 = numberReg)
                     it += IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=zeroReg, immediate = 0)
-                    it += IRInstruction(compareAndSetOpcode, IRDataType.BYTE, reg1=cmpReg, reg2=zeroReg)
-                    it += IRInstruction(Opcode.FFROMUB, IRDataType.FLOAT, reg1=cmpReg, fpReg1 = valueReg)
+                    it += IRInstruction(compareAndSetOpcode, IRDataType.BYTE, reg1=cmpResultReg, reg2=cmpReg, reg3=zeroReg)
+                    it += IRInstruction(Opcode.FFROMUB, IRDataType.FLOAT, reg1=cmpResultReg, fpReg1 = valueReg)
                     it += IRInstruction(Opcode.STOREM, IRDataType.FLOAT, fpReg1 = valueReg, address = knownAddress)
                 }
             } else {
@@ -1311,13 +1315,14 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                     it += IRInstruction(Opcode.LOAD, IRDataType.FLOAT, fpReg1 = numberReg, immediateFp = operand.number.toFloat())
                     it += IRInstruction(Opcode.FCOMP, IRDataType.FLOAT, reg1=cmpReg, fpReg1 = valueReg, fpReg2 = numberReg)
                     it += IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=zeroReg, immediate = 0)
-                    it += IRInstruction(compareAndSetOpcode, IRDataType.BYTE, reg1=cmpReg, reg2=zeroReg)
-                    it += IRInstruction(Opcode.FFROMUB, IRDataType.FLOAT, reg1=cmpReg, fpReg1 = valueReg)
+                    it += IRInstruction(compareAndSetOpcode, IRDataType.BYTE, reg1=cmpResultReg, reg2=cmpReg, reg3=zeroReg)
+                    it += IRInstruction(Opcode.FFROMUB, IRDataType.FLOAT, reg1=cmpResultReg, fpReg1 = valueReg)
                     it += IRInstruction(Opcode.STOREM, IRDataType.FLOAT, fpReg1 = valueReg, labelSymbol = symbol)
                 }
             }
         } else {
             val tr = translateExpression(operand)
+            val cmpResultReg = codeGen.registers.nextFree()
             addToResult(result, tr, -1, tr.resultFpReg)
             if (knownAddress != null) {
                 // in-place modify a memory location
@@ -1325,8 +1330,8 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                     it += IRInstruction(Opcode.LOADM, IRDataType.FLOAT, fpReg1 = valueReg, address = knownAddress)
                     it += IRInstruction(Opcode.FCOMP, IRDataType.FLOAT, reg1=cmpReg, fpReg1 = valueReg, fpReg2 = tr.resultFpReg)
                     it += IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=zeroReg, immediate = 0)
-                    it += IRInstruction(compareAndSetOpcode, IRDataType.BYTE, reg1=cmpReg, reg2=zeroReg)
-                    it += IRInstruction(Opcode.FFROMUB, IRDataType.FLOAT, reg1=cmpReg, fpReg1 = valueReg)
+                    it += IRInstruction(compareAndSetOpcode, IRDataType.BYTE, reg1=cmpResultReg, reg2=cmpReg, reg3=zeroReg)
+                    it += IRInstruction(Opcode.FFROMUB, IRDataType.FLOAT, reg1=cmpResultReg, fpReg1 = valueReg)
                     it += IRInstruction(Opcode.STOREM, IRDataType.FLOAT, fpReg1 = valueReg, address = knownAddress)
                 }
             } else {
@@ -1335,8 +1340,8 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                     it += IRInstruction(Opcode.LOADM, IRDataType.FLOAT, fpReg1 = valueReg, labelSymbol = symbol)
                     it += IRInstruction(Opcode.FCOMP, IRDataType.FLOAT, reg1=cmpReg, fpReg1 = valueReg, fpReg2 = tr.resultFpReg)
                     it += IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=zeroReg, immediate = 0)
-                    it += IRInstruction(compareAndSetOpcode, IRDataType.BYTE, reg1=cmpReg, reg2=zeroReg)
-                    it += IRInstruction(Opcode.FFROMUB, IRDataType.FLOAT, reg1=cmpReg, fpReg1 = valueReg)
+                    it += IRInstruction(compareAndSetOpcode, IRDataType.BYTE, reg1=cmpResultReg, reg2=cmpReg, reg3=zeroReg)
+                    it += IRInstruction(Opcode.FFROMUB, IRDataType.FLOAT, reg1=cmpResultReg, fpReg1 = valueReg)
                     it += IRInstruction(Opcode.STOREM, IRDataType.FLOAT, fpReg1 = valueReg, labelSymbol = symbol)
                 }
             }

codeGenIntermediate/src/prog8/codegen/intermediate/IRCodeGen.kt

@@ -1,9 +1,6 @@
 package prog8.codegen.intermediate
 
-import prog8.code.StMemVar
-import prog8.code.StNode
-import prog8.code.StStaticVariable
-import prog8.code.SymbolTable
+import prog8.code.*
 import prog8.code.ast.*
 import prog8.code.core.*
 import prog8.intermediate.*
@@ -32,9 +29,6 @@ class IRCodeGen(
         irSymbolTable = IRSymbolTable(symbolTable)
         val irProg = IRProgram(program.name, irSymbolTable, options, program.encoding)
 
-        if(options.evalStackBaseAddress!=null)
-            throw AssemblyError("IR doesn't use eval-stack")
-
         // collect global variables initializers
         program.allBlocks().forEach {
             val result = mutableListOf<IRCodeChunkBase>()
@@ -102,14 +96,16 @@ class IRCodeGen(
         // make sure that first chunks in Blocks and Subroutines share the name of the block/sub as label.
 
         irProg.blocks.forEach { block ->
-            block.children.firstOrNull { it is IRInlineAsmChunk }?.let { first->
-                first as IRInlineAsmChunk
-                if(first.label==null) {
-                    val replacement = IRInlineAsmChunk(block.label, first.assembly, first.isIR, first.next)
-                    block.children.removeAt(0)
-                    block.children.add(0, replacement)
-                } else if(first.label != block.label) {
-                    throw AssemblyError("first chunk in block has label that differs from block name")
+            if(block.isNotEmpty()) {
+                val firstAsm = block.children[0] as? IRInlineAsmChunk
+                if(firstAsm!=null) {
+                    if(firstAsm.label==null) {
+                        val replacement = IRInlineAsmChunk(block.label, firstAsm.assembly, firstAsm.isIR, firstAsm.next)
+                        block.children.removeAt(0)
+                        block.children.add(0, replacement)
+                    } else if(firstAsm.label != block.label) {
+                        throw AssemblyError("first chunk in block has label that differs from block name")
+                    }
                 }
             }
 
@@ -168,16 +164,22 @@ class IRCodeGen(
 
         replacements.forEach {
             val old = it.first.instructions[it.second]
+            val formats = instructionFormats.getValue(old.opcode)
+            val format = formats.getOrElse(old.type) { throw IllegalArgumentException("type ${old.type} invalid for ${old.opcode}") }
+            val immediateValue = if(format.immediate) it.third.toInt() else null
+            val addressValue = if(format.immediate) null else it.third.toInt()
+
             it.first.instructions[it.second] = IRInstruction(
                 old.opcode,
                 old.type,
                 old.reg1,
                 old.reg2,
+                old.reg3,
                 old.fpReg1,
                 old.fpReg2,
+                immediate = immediateValue,
                 null,
-                null,
-                address = it.third.toInt(),
+                address = addressValue,
                 null,
                 null
             )
@@ -287,6 +289,8 @@ class IRCodeGen(
             }
         }
 
+        chunks.filterIsInstance<IRCodeChunk>().firstOrNull()?.appendSrcPosition(node.position)
+
         return chunks
     }
 
@@ -395,42 +399,42 @@ class IRCodeGen(
     }
 
     private fun translate(whenStmt: PtWhen): IRCodeChunks {
-        if(whenStmt.choices.children.isEmpty())
-            return emptyList()
         val result = mutableListOf<IRCodeChunkBase>()
         val valueDt = irType(whenStmt.value.type)
         val valueTr = expressionEval.translateExpression(whenStmt.value)
         addToResult(result, valueTr, valueTr.resultReg, -1)
-        val choices = whenStmt.choices.children.map {it as PtWhenChoice }
+
+        val choices = mutableListOf<Pair<String, PtWhenChoice>>()
         val endLabel = createLabelName()
-        for (choice in choices) {
+        whenStmt.choices.children.forEach {
+            val choice = it as PtWhenChoice
             if(choice.isElse) {
                 result += translateNode(choice.statements)
+                addInstr(result, IRInstruction(Opcode.JUMP, labelSymbol = endLabel), null)
             } else {
-                val skipLabel = createLabelName()
-                val values = choice.values.children.map {it as PtNumber}
-                if(values.size==1) {
-                    val chunk = IRCodeChunk(null, null)
-                    chunk += IRInstruction(Opcode.BNE, valueDt, reg1=valueTr.resultReg, immediate = values[0].number.toInt(), labelSymbol = skipLabel)
-                    result += chunk
-                    result += translateNode(choice.statements)
-                    if(choice.statements.children.last() !is PtReturn)
-                        addInstr(result, IRInstruction(Opcode.JUMP, labelSymbol = endLabel), null)
+                if(choice.statements.children.isEmpty()) {
+                    // no statements for this choice value, jump to the end immediately
+                    choice.values.children.map { it as PtNumber }.sortedBy { it.number }.forEach { value ->
+                        addInstr(result, IRInstruction(Opcode.BEQ, valueDt, reg1=valueTr.resultReg, immediate = value.number.toInt(), labelSymbol = endLabel), null)
+                    }
                 } else {
-                    val matchLabel = createLabelName()
-                    val chunk = IRCodeChunk(null, null)
-                    for (value in values) {
-                        chunk += IRInstruction(Opcode.BEQ, valueDt, reg1=valueTr.resultReg, immediate = value.number.toInt(), labelSymbol = matchLabel)
+                    val choiceLabel = createLabelName()
+                    choices.add(choiceLabel to choice)
+                    choice.values.children.map { it as PtNumber }.sortedBy { it.number }.forEach { value ->
+                        addInstr(result, IRInstruction(Opcode.BEQ, valueDt, reg1=valueTr.resultReg, immediate = value.number.toInt(), labelSymbol = choiceLabel), null)
                     }
-                    chunk += IRInstruction(Opcode.JUMP, labelSymbol = skipLabel)
-                    result += chunk
-                    result += labelFirstChunk(translateNode(choice.statements), matchLabel)
-                    if(choice.statements.children.last() !is PtReturn)
-                        addInstr(result, IRInstruction(Opcode.JUMP, labelSymbol = endLabel), null)
                 }
-                result += IRCodeChunk(skipLabel, null)
             }
         }
+        addInstr(result, IRInstruction(Opcode.JUMP, labelSymbol = endLabel), null)
+
+        choices.forEach { (label, choice) ->
+            result += labelFirstChunk(translateNode(choice.statements), label)
+            val lastStatement = choice.statements.children.last()
+            if(lastStatement !is PtReturn && lastStatement !is PtJump)
+                addInstr(result, IRInstruction(Opcode.JUMP, labelSymbol = endLabel), null)
+        }
+
         result += IRCodeChunk(endLabel, null)
         return result
     }
@@ -447,60 +451,64 @@ class IRCodeGen(
                     result += translateForInNonConstantRange(forLoop, loopvar)
             }
             is PtIdentifier -> {
-                val iterableVar = symbolTable.lookup(iterable.name) as StStaticVariable
                 require(forLoop.variable.name == loopvar.scopedName)
+                val iterableLength = symbolTable.getLength(iterable.name)
                 val loopvarSymbol = forLoop.variable.name
                 val indexReg = registers.nextFree()
                 val tmpReg = registers.nextFree()
                 val loopLabel = createLabelName()
                 val endLabel = createLabelName()
-                if(iterableVar.dt==DataType.STR) {
-                    // iterate over a zero-terminated string
-                    addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = indexReg, immediate = 0), null)
-                    result += IRCodeChunk(loopLabel, null).also {
-                        it += IRInstruction(Opcode.LOADX, IRDataType.BYTE, reg1 = tmpReg, reg2 = indexReg, labelSymbol = iterable.name)
-                        it += IRInstruction(Opcode.BEQ, IRDataType.BYTE, reg1 = tmpReg, immediate = 0, labelSymbol = endLabel)
-                        it += IRInstruction(Opcode.STOREM, IRDataType.BYTE, reg1 = tmpReg, labelSymbol = loopvarSymbol)
+                when (iterable.type) {
+                    DataType.STR -> {
+                        // iterate over a zero-terminated string
+                        addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = indexReg, immediate = 0), null)
+                        result += IRCodeChunk(loopLabel, null).also {
+                            it += IRInstruction(Opcode.LOADX, IRDataType.BYTE, reg1 = tmpReg, reg2 = indexReg, labelSymbol = iterable.name)
+                            it += IRInstruction(Opcode.BEQ, IRDataType.BYTE, reg1 = tmpReg, immediate = 0, labelSymbol = endLabel)
+                            it += IRInstruction(Opcode.STOREM, IRDataType.BYTE, reg1 = tmpReg, labelSymbol = loopvarSymbol)
+                        }
+                        result += translateNode(forLoop.statements)
+                        val jumpChunk = IRCodeChunk(null, null)
+                        jumpChunk += IRInstruction(Opcode.INC, IRDataType.BYTE, reg1 = indexReg)
+                        jumpChunk += IRInstruction(Opcode.JUMP, labelSymbol = loopLabel)
+                        result += jumpChunk
+                        result += IRCodeChunk(endLabel, null)
                     }
-                    result += translateNode(forLoop.statements)
-                    val jumpChunk = IRCodeChunk(null, null)
-                    jumpChunk += IRInstruction(Opcode.INC, IRDataType.BYTE, reg1 = indexReg)
-                    jumpChunk += IRInstruction(Opcode.JUMP, labelSymbol = loopLabel)
-                    result += jumpChunk
-                    result += IRCodeChunk(endLabel, null)
-                } else if(iterable.type in SplitWordArrayTypes) {
-                    // iterate over lsb/msb split word array
-                    val elementDt = ArrayToElementTypes.getValue(iterable.type)
-                    if(elementDt !in WordDatatypes)
-                        throw AssemblyError("weird dt")
-                    val length = iterableVar.length!!
-                    addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=indexReg, immediate = 0), null)
-                    result += IRCodeChunk(loopLabel, null).also {
-                        val tmpRegLsb = registers.nextFree()
-                        val tmpRegMsb = registers.nextFree()
-                        it += IRInstruction(Opcode.LOADX, IRDataType.BYTE, reg1=tmpRegLsb, reg2=indexReg, immediate = length, labelSymbol=iterable.name+"_lsb")
-                        it += IRInstruction(Opcode.LOADX, IRDataType.BYTE, reg1=tmpRegMsb, reg2=indexReg, immediate = length, labelSymbol=iterable.name+"_msb")
-                        it += IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=tmpRegLsb, reg2=tmpRegMsb)
-                        it += IRInstruction(Opcode.STOREM, irType(elementDt), reg1=tmpRegLsb, labelSymbol = loopvarSymbol)
+                    in SplitWordArrayTypes -> {
+                        // iterate over lsb/msb split word array
+                        val elementDt = ArrayToElementTypes.getValue(iterable.type)
+                        if(elementDt !in WordDatatypes)
+                            throw AssemblyError("weird dt")
+                        addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=indexReg, immediate = 0), null)
+                        result += IRCodeChunk(loopLabel, null).also {
+                            val tmpRegLsb = registers.nextFree()
+                            val tmpRegMsb = registers.nextFree()
+                            val concatReg = registers.nextFree()
+                            it += IRInstruction(Opcode.LOADX, IRDataType.BYTE, reg1=tmpRegLsb, reg2=indexReg, immediate = iterableLength, labelSymbol=iterable.name+"_lsb")
+                            it += IRInstruction(Opcode.LOADX, IRDataType.BYTE, reg1=tmpRegMsb, reg2=indexReg, immediate = iterableLength, labelSymbol=iterable.name+"_msb")
+                            it += IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=concatReg, reg2=tmpRegLsb, reg3=tmpRegMsb)
+                            it += IRInstruction(Opcode.STOREM, irType(elementDt), reg1=concatReg, labelSymbol = loopvarSymbol)
+                        }
+                        result += translateNode(forLoop.statements)
+                        result += IRCodeChunk(null, null).also {
+                            it += IRInstruction(Opcode.INC, IRDataType.BYTE, reg1=indexReg)
+                            it += IRInstruction(Opcode.BNE, IRDataType.BYTE, reg1=indexReg, immediate = if(iterableLength==256) 0 else iterableLength, labelSymbol = loopLabel)
+                        }
                     }
-                    result += translateNode(forLoop.statements)
-                    result += IRCodeChunk(null, null).also {
-                        it += IRInstruction(Opcode.INC, IRDataType.BYTE, reg1=indexReg)
-                        it += IRInstruction(Opcode.BNE, IRDataType.BYTE, reg1=indexReg, immediate = if(length==256) 0 else length, labelSymbol = loopLabel)
+                    else -> {
+                        // iterate over regular array
+                        val elementDt = ArrayToElementTypes.getValue(iterable.type)
+                        val elementSize = program.memsizer.memorySize(elementDt)
+                        val lengthBytes = iterableLength!! * elementSize
+                        addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=indexReg, immediate = 0), null)
+                        result += IRCodeChunk(loopLabel, null).also {
+                            it += IRInstruction(Opcode.LOADX, irType(elementDt), reg1=tmpReg, reg2=indexReg, labelSymbol=iterable.name)
+                            it += IRInstruction(Opcode.STOREM, irType(elementDt), reg1=tmpReg, labelSymbol = loopvarSymbol)
+                        }
+                        result += translateNode(forLoop.statements)
+                        result += addConstReg(IRDataType.BYTE, indexReg, elementSize)
+                        addInstr(result, IRInstruction(Opcode.BNE, IRDataType.BYTE, reg1=indexReg, immediate = if(lengthBytes==256) 0 else lengthBytes, labelSymbol = loopLabel), null)
                     }
-                } else {
-                    // iterate over regular array
-                    val elementDt = ArrayToElementTypes.getValue(iterable.type)
-                    val elementSize = program.memsizer.memorySize(elementDt)
-                    val lengthBytes = iterableVar.length!! * elementSize
-                    addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=indexReg, immediate = 0), null)
-                    result += IRCodeChunk(loopLabel, null).also {
-                        it += IRInstruction(Opcode.LOADX, irType(elementDt), reg1=tmpReg, reg2=indexReg, labelSymbol=iterable.name)
-                        it += IRInstruction(Opcode.STOREM, irType(elementDt), reg1=tmpReg, labelSymbol = loopvarSymbol)
-                    }
-                    result += translateNode(forLoop.statements)
-                    result += addConstReg(IRDataType.BYTE, indexReg, elementSize)
-                    addInstr(result, IRInstruction(Opcode.BNE, IRDataType.BYTE, reg1=indexReg, immediate = if(lengthBytes==256) 0 else lengthBytes, labelSymbol = loopLabel), null)
                 }
             }
             else -> throw AssemblyError("weird for iterable")
@@ -545,22 +553,24 @@ class IRCodeGen(
 
         addInstr(result, IRInstruction(Opcode.STOREM, loopvarDtIr, reg1=fromTr.resultReg, labelSymbol=loopvarSymbol), null)
         result += labelFirstChunk(translateNode(forLoop.statements), loopLabel)
-        result += addConstMem(loopvarDtIr, null, loopvarSymbol, step)
-        addInstr(result, IRInstruction(Opcode.LOADM, loopvarDtIr, reg1 = fromTr.resultReg, labelSymbol = loopvarSymbol), null)
-        // if endvalue >= index, iterate loop
-        val branchInstr = if(loopvarDt in SignedDatatypes) {
-            if(step>0)
-                IRInstruction(Opcode.BGESR, loopvarDtIr, reg1=toTr.resultReg, reg2=fromTr.resultReg, labelSymbol=loopLabel)
-            else
-                IRInstruction(Opcode.BGESR, loopvarDtIr, reg1=fromTr.resultReg, reg2=toTr.resultReg, labelSymbol=loopLabel)
+        if(step==1 || step==-1) {
+            // if endvalue == loopvar, stop loop, else iterate
+            addInstr(result, IRInstruction(Opcode.LOADM, loopvarDtIr, reg1 = fromTr.resultReg, labelSymbol = loopvarSymbol), null)
+            addInstr(result, IRInstruction(Opcode.BEQR, loopvarDtIr, reg1=toTr.resultReg, reg2=fromTr.resultReg, labelSymbol = labelAfterFor), null)
+            result += addConstMem(loopvarDtIr, null, loopvarSymbol, step)
+            addInstr(result, IRInstruction(Opcode.JUMP, labelSymbol = loopLabel), null)
         } else {
-            if(step>0)
-                IRInstruction(Opcode.BGER, loopvarDtIr, reg1=toTr.resultReg, reg2=fromTr.resultReg, labelSymbol=loopLabel)
+            // ind/dec index, then:
+            // ascending: if endvalue >= loopvar, iterate
+            // descending: if loopvar >= endvalue, iterate
+            result += addConstMem(loopvarDtIr, null, loopvarSymbol, step)
+            addInstr(result, IRInstruction(Opcode.LOADM, loopvarDtIr, reg1=fromTr.resultReg, labelSymbol = loopvarSymbol), null)
+            if(step > 0)
+                addInstr(result, IRInstruction(Opcode.CMP, loopvarDtIr, reg1 = toTr.resultReg, fromTr.resultReg), null)
             else
-                IRInstruction(Opcode.BGER, loopvarDtIr, reg1=fromTr.resultReg, reg2=toTr.resultReg, labelSymbol=loopLabel)
+                addInstr(result, IRInstruction(Opcode.CMP, loopvarDtIr, reg1 = fromTr.resultReg, toTr.resultReg), null)
+            addInstr(result, IRInstruction(Opcode.BSTPOS, labelSymbol = loopLabel), null)
         }
-        addInstr(result, branchInstr, null)
-
         result += IRCodeChunk(labelAfterFor, null)
         return result
     }
@@ -576,25 +586,23 @@ class IRCodeGen(
             else -> throw AssemblyError("invalid loopvar node type")
         }
         val loopvarDtIr = irType(loopvarDt)
-        val iterable = forLoop.iterable as PtRange
-        val step = iterable.step.number.toInt()
-        val rangeStart = (iterable.from as PtNumber).number.toInt()
-        val rangeEndUntyped = (iterable.to as PtNumber).number.toInt() + step
-        if(step==0)
-            throw AssemblyError("step 0")
-        if(step>0 && rangeEndUntyped<rangeStart || step<0 && rangeEndUntyped>rangeStart)
+        val iterable = (forLoop.iterable as PtRange).toConstantIntegerRange()!!
+        if(iterable.isEmpty())
             throw AssemblyError("empty range")
-        val rangeEndWrapped = if(loopvarDtIr==IRDataType.BYTE) rangeEndUntyped and 255 else rangeEndUntyped and 65535
+        if(iterable.step==0)
+            throw AssemblyError("step 0")
+        val rangeEndExclusiveUntyped = iterable.last + iterable.step
+        val rangeEndExclusiveWrapped = if(loopvarDtIr==IRDataType.BYTE) rangeEndExclusiveUntyped and 255 else rangeEndExclusiveUntyped and 65535
         val result = mutableListOf<IRCodeChunkBase>()
         val chunk = IRCodeChunk(null, null)
-        chunk += IRInstruction(Opcode.LOAD, loopvarDtIr, reg1=indexReg, immediate = rangeStart)
+        chunk += IRInstruction(Opcode.LOAD, loopvarDtIr, reg1=indexReg, immediate = iterable.first)
         chunk += IRInstruction(Opcode.STOREM, loopvarDtIr, reg1=indexReg, labelSymbol=loopvarSymbol)
         result += chunk
         result += labelFirstChunk(translateNode(forLoop.statements), loopLabel)
-        result += addConstMem(loopvarDtIr, null, loopvarSymbol, step)
-        val chunk2 = IRCodeChunk(null, null)
+        val chunk2 = addConstMem(loopvarDtIr, null, loopvarSymbol, iterable.step)
         chunk2 += IRInstruction(Opcode.LOADM, loopvarDtIr, reg1 = indexReg, labelSymbol = loopvarSymbol)
-        chunk2 += IRInstruction(Opcode.BNE, loopvarDtIr, reg1 = indexReg, immediate = rangeEndWrapped, labelSymbol = loopLabel)
+        chunk2 += IRInstruction(Opcode.XOR, loopvarDtIr, reg1 = 999, immediate = 111)
+        chunk2 += IRInstruction(Opcode.BNE, loopvarDtIr, reg1 = indexReg, immediate = rangeEndExclusiveWrapped, labelSymbol = loopLabel)
         result += chunk2
         return result
     }
@@ -905,7 +913,6 @@ class IRCodeGen(
         val goto = ifElse.ifScope.children.firstOrNull() as? PtJump
         when (condition) {
             is PtBinaryExpression -> {
-                require(!options.useNewExprCode)
                 if(condition.operator !in ComparisonOperators)
                     throw AssemblyError("if condition should only be a binary comparison expression")
 
@@ -919,7 +926,6 @@ class IRCodeGen(
             }
             else -> {
                 // if X   --> meaning:  if X!=0
-                require(options.useNewExprCode)
                 val irDt = irType(condition.type)
                 val signed = condition.type in SignedDatatypes
                 return if(goto!=null && ifElse.elseScope.children.isEmpty()) {
@@ -997,10 +1003,10 @@ class IRCodeGen(
         val opcode = when (condition.operator) {
             "==" -> Opcode.BEQ
             "!=" -> Opcode.BNE
-            "<" -> if (signed) Opcode.BLTS else throw AssemblyError("unsigned < 0 shouldn't occur in codegen")
-            ">" -> if (signed) Opcode.BGTS else throw AssemblyError("unsigned > 0 shouldn't occur in codegen")
-            "<=" -> if (signed) Opcode.BLES else throw AssemblyError("unsigned <= 0 shouldn't occur in codegen")
-            ">=" -> if (signed) Opcode.BGES else throw AssemblyError("unsigned >= 0 shouldn't occur in codegen")
+            "<" -> if (signed) Opcode.BLTS else Opcode.BLT
+            ">" -> if (signed) Opcode.BGTS else Opcode.BGT
+            "<=" -> if (signed) Opcode.BLES else Opcode.BLE
+            ">=" -> if (signed) Opcode.BGES else Opcode.BGE
             else -> throw AssemblyError("invalid comparison operator")
         }
         if (goto.address != null)
@@ -1134,10 +1140,10 @@ class IRCodeGen(
             elseBranch = when (condition.operator) {
                 "==" -> Opcode.BNE
                 "!=" -> Opcode.BEQ
-                "<" -> if (signed) Opcode.BGES else throw AssemblyError("unsigned < 0 shouldn't occur in codegen")
-                ">" -> if (signed) Opcode.BLES else throw AssemblyError("unsigned > 0 shouldn't occur in codegen")
-                "<=" -> if (signed) Opcode.BGTS else throw AssemblyError("unsigned <= 0 shouldn't occur in codegen")
-                ">=" -> if (signed) Opcode.BLTS else throw AssemblyError("unsigned >= 0 shouldn't occur in codegen")
+                "<" -> if (signed) Opcode.BGES else Opcode.BGE
+                ">" -> if (signed) Opcode.BLES else Opcode.BLE
+                "<=" -> if (signed) Opcode.BGTS else Opcode.BGT
+                ">=" -> if (signed) Opcode.BLTS else Opcode.BLT
                 else -> throw AssemblyError("weird operator")
             }
         }
@@ -1364,7 +1370,6 @@ class IRCodeGen(
         if(array?.splitWords==true) {
             val variable = array.variable.name
             val fixedIndex = constIntValue(array.index)
-            val iterable = symbolTable.flat.getValue(array.variable.name) as StStaticVariable
             if(fixedIndex!=null) {
                 val skipLabel = createLabelName()
                 when(postIncrDecr.operator) {
@@ -1390,7 +1395,7 @@ class IRCodeGen(
                     else -> throw AssemblyError("weird operator")
                 }
             } else {
-                val arrayLength = iterable.length!!
+                val arrayLength = symbolTable.getLength(array.variable.name)
                 val indexTr = expressionEval.translateExpression(array.index)
                 addToResult(result, indexTr, indexTr.resultReg, -1)
                 val incReg = registers.nextFree()
@@ -1464,7 +1469,19 @@ class IRCodeGen(
                 val fixedIndex = constIntValue(array.index)
                 if(fixedIndex!=null) {
                     val offset = fixedIndex*itemsize
-                    addInstr(result, IRInstruction(operationMem, irDt, labelSymbol="$variable+$offset"), null)
+                    val indexReg = registers.nextFree()
+                    val dataReg = registers.nextFree()
+                    if(array.usesPointerVariable) {
+                        // we don't have an indirect dec/inc so do it via an intermediate register
+                        result += IRCodeChunk(null,null).also {
+                            it += IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = indexReg, immediate = offset)
+                            it += IRInstruction(Opcode.LOADIX, irDt, reg1 = dataReg, reg2 = indexReg, labelSymbol = variable)
+                            it += IRInstruction(operationRegister, irDt, reg1=dataReg)
+                            it += IRInstruction(Opcode.STOREIX, irDt, reg1 = dataReg, reg2 = indexReg, labelSymbol = variable)
+                        }
+                    } else {
+                        addInstr(result, IRInstruction(operationMem, irDt, labelSymbol = "$variable+$offset"), null)
+                    }
                 } else {
                     val indexTr = expressionEval.translateExpression(array.index)
                     addToResult(result, indexTr, indexTr.resultReg, -1)
@@ -1472,9 +1489,15 @@ class IRCodeGen(
                         result += multiplyByConst(IRDataType.BYTE, indexTr.resultReg, itemsize)
                     result += IRCodeChunk(null, null).also {
                         val incReg = registers.nextFree()
-                        it += IRInstruction(Opcode.LOADX, irDt, reg1=incReg, reg2=indexTr.resultReg, labelSymbol=variable)
-                        it += IRInstruction(operationRegister, irDt, reg1=incReg)
-                        it += IRInstruction(Opcode.STOREX, irDt, reg1=incReg, reg2=indexTr.resultReg, labelSymbol=variable)
+                        if(array.usesPointerVariable) {
+                            it += IRInstruction(Opcode.LOADIX, irDt, reg1=incReg, reg2=indexTr.resultReg, labelSymbol=variable)
+                            it += IRInstruction(operationRegister, irDt, reg1=incReg)
+                            it += IRInstruction(Opcode.STOREIX, irDt, reg1=incReg, reg2=indexTr.resultReg, labelSymbol=variable)
+                        } else {
+                            it += IRInstruction(Opcode.LOADX, irDt, reg1=incReg, reg2=indexTr.resultReg, labelSymbol=variable)
+                            it += IRInstruction(operationRegister, irDt, reg1=incReg)
+                            it += IRInstruction(Opcode.STOREX, irDt, reg1=incReg, reg2=indexTr.resultReg, labelSymbol=variable)
+                        }
                     }
                 }
             } else
@@ -1500,7 +1523,10 @@ class IRCodeGen(
         val result = mutableListOf<IRCodeChunkBase>()
         val countTr = expressionEval.translateExpression(repeat.count)
         addToResult(result, countTr, countTr.resultReg, -1)
-        addInstr(result, IRInstruction(Opcode.BEQ, irDt, reg1=countTr.resultReg, immediate = 0, labelSymbol = skipRepeatLabel), null)
+        if(constIntValue(repeat.count)==null) {
+            // check if the counter is already zero
+            addInstr(result, IRInstruction(Opcode.BEQ, irDt, reg1=countTr.resultReg, immediate = 0, labelSymbol = skipRepeatLabel), null)
+        }
         result += labelFirstChunk(translateNode(repeat.statements), repeatLabel)
         result += IRCodeChunk(null, null).also {
             it += IRInstruction(Opcode.DEC, irDt, reg1 = countTr.resultReg)
@@ -1512,17 +1538,26 @@ class IRCodeGen(
 
     private fun translate(jump: PtJump): IRCodeChunks {
         val result = mutableListOf<IRCodeChunkBase>()
-        val instr = if(jump.address!=null) {
-            IRInstruction(Opcode.JUMPA, address = jump.address!!.toInt())
+        val chunk = IRCodeChunk(null, null)
+        if(jump.address!=null) {
+            chunk += IRInstruction(Opcode.JUMP, address = jump.address!!.toInt())
         } else {
             if (jump.generatedLabel != null)
-                IRInstruction(Opcode.JUMP, labelSymbol = jump.generatedLabel!!)
-            else if (jump.identifier != null)
-                IRInstruction(Opcode.JUMP, labelSymbol = jump.identifier!!.name)
+                chunk += IRInstruction(Opcode.JUMP, labelSymbol = jump.generatedLabel!!)
+            else if (jump.identifier != null) {
+                val symbol = symbolTable.lookup(jump.identifier!!.name)
+                if(symbol?.type==StNodeType.MEMVAR || symbol?.type==StNodeType.STATICVAR) {
+                    val jumpReg = registers.nextFree()
+                    chunk += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1 = jumpReg, labelSymbol = jump.identifier!!.name)
+                    chunk += IRInstruction(Opcode.JUMPI, reg1 = jumpReg)
+                } else {
+                    chunk += IRInstruction(Opcode.JUMP, labelSymbol = jump.identifier!!.name)
+                }
+            }
             else
                 throw AssemblyError("weird jump")
         }
-        addInstr(result, instr, null)
+        result += chunk
         return result
     }
 
@@ -1610,7 +1645,7 @@ class IRCodeGen(
     private fun translate(parameters: List<PtSubroutineParameter>) =
         parameters.map {
             val flattenedName = it.definingSub()!!.name + "." + it.name
-            val orig = symbolTable.flat.getValue(flattenedName) as StStaticVariable
+            val orig = symbolTable.lookup(flattenedName) as StStaticVariable
             IRSubroutine.IRParam(flattenedName, orig.dt)
         }
 
@@ -1635,24 +1670,6 @@ class IRCodeGen(
 
     internal fun isOne(expression: PtExpression): Boolean = expression is PtNumber && expression.number==1.0
 
-    fun getReusableTempvar(scope: PtNamedNode, type: DataType): PtIdentifier {
-        val uniqueId = Pair(scope, type).hashCode().toUInt()
-        val tempvarname = "${scope.scopedName}.tempvar_${uniqueId}"
-        val tempvar = PtIdentifier(tempvarname, type, Position.DUMMY)
-        val staticVar = StStaticVariable(
-            tempvarname,
-            type,
-            null,
-            null,
-            null,
-            null,
-            ZeropageWish.DONTCARE,
-            tempvar
-        )
-        irSymbolTable.add(staticVar)
-        return tempvar
-    }
-
     fun makeSyscall(syscall: IMSyscall, params: List<Pair<IRDataType, Int>>, returns: Pair<IRDataType, Int>?, label: String?=null): IRCodeChunk {
         return IRCodeChunk(label, null).also {
             val args = params.map { (dt, reg)->

codeGenIntermediate/src/prog8/codegen/intermediate/IRPeepholeOptimizer.kt

@@ -37,8 +37,6 @@ class IRPeepholeOptimizer(private val irprog: IRProgram) {
             removeEmptyChunks(sub)
             joinChunks(sub)
 
-            // TODO also do register optimization step here?
-
             sub.chunks.withIndex().forEach { (index, chunk1) ->
                 // we don't optimize Inline Asm chunks here.
                 val chunk2 = if(index<sub.chunks.size-1) sub.chunks[index+1] else null
@@ -49,17 +47,19 @@ class IRPeepholeOptimizer(private val irprog: IRProgram) {
                         val changed = removeNops(chunk1, indexedInstructions)
                                 || removeDoubleLoadsAndStores(chunk1, indexedInstructions)       // TODO not yet implemented
                                 || removeUselessArithmetic(chunk1, indexedInstructions)
+                                || removeNeedlessCompares(chunk1, indexedInstructions)
                                 || removeWeirdBranches(chunk1, chunk2, indexedInstructions)
                                 || removeDoubleSecClc(chunk1, indexedInstructions)
                                 || cleanupPushPop(chunk1, indexedInstructions)
-                        // TODO other optimizations:
-                        //  more complex optimizations such as unused registers
+                        // TODO other optimizations
                     } while (changed)
                 }
             }
             removeEmptyChunks(sub)
         }
 
+        // TODO also do register optimization step here at the end?
+
         irprog.linkChunks()  // re-link
     }
 
@@ -105,7 +105,10 @@ class IRPeepholeOptimizer(private val irprog: IRProgram) {
 
         relabelChunks.forEach { (index, label) ->
             val chunk = IRCodeChunk(label, null)
-            chunk.instructions += sub.chunks[index].instructions
+            val subChunk = sub.chunks[index]
+            chunk.instructions += subChunk.instructions
+            if(subChunk is IRCodeChunk)
+                chunk.appendSrcPositions(subChunk.sourceLinesPositions)
             sub.chunks[index] = chunk
         }
         removeChunks.reversed().forEach { index -> sub.chunks.removeAt(index) }
@@ -140,6 +143,8 @@ class IRPeepholeOptimizer(private val irprog: IRProgram) {
                     if(mayJoinCodeChunks(lastChunk, candidate)) {
                         lastChunk.instructions += candidate.instructions
                         lastChunk.next = candidate.next
+                        if(lastChunk is IRCodeChunk)
+                            lastChunk.appendSrcPositions(candidate.sourceLinesPositions)
                     }
                     else
                         chunks += candidate
@@ -267,6 +272,28 @@ class IRPeepholeOptimizer(private val irprog: IRProgram) {
         return changed
     }
 
+    private fun removeNeedlessCompares(chunk: IRCodeChunk, indexedInstructions: List<IndexedValue<IRInstruction>>): Boolean {
+        // a CMPI with 0, after an instruction like LOAD that already sets the status bits, can be removed.
+        // but only if the instruction after it is not using the Carry bit because that won't be set by a LOAD instruction etc.
+        var changed = false
+        indexedInstructions.reversed().forEach { (idx, ins) ->
+            if(idx>0 && idx<(indexedInstructions.size-1) && ins.opcode==Opcode.CMPI && ins.immediate==0) {
+                val previous = indexedInstructions[idx-1].value
+                if(previous.opcode in OpcodesThatSetStatusbitsIncludingCarry) {
+                    chunk.instructions.removeAt(idx)
+                    changed = true
+                } else if(previous.opcode in OpcodesThatSetStatusbitsButNotCarry) {
+                    val next = indexedInstructions[idx+1].value
+                    if(next.opcode !in arrayOf(Opcode.BSTCC, Opcode.BSTCS, Opcode.BSTPOS, Opcode.BSTNEG)) {
+                        chunk.instructions.removeAt(idx)
+                        changed = true
+                    }
+                }
+            }
+        }
+        return changed
+    }
+
     private fun removeUselessArithmetic(chunk: IRCodeChunk, indexedInstructions: List<IndexedValue<IRInstruction>>): Boolean {
         // note: this is hard to solve for the non-immediate instructions atm because the values are loaded into registers first
         var changed = false

codeGenIntermediate/src/prog8/codegen/intermediate/IRUnusedCodeRemover.kt

@@ -1,7 +1,7 @@
 package prog8.codegen.intermediate
 
 import prog8.code.core.IErrorReporter
-import prog8.code.core.SourceCode.Companion.libraryFilePrefix
+import prog8.code.core.SourceCode.Companion.LIBRARYFILEPREFIX
 import prog8.intermediate.*
 
 
@@ -16,7 +16,7 @@ class IRUnusedCodeRemover(
         irprog.blocks.reversed().forEach { block ->
             if(block.isEmpty()) {
                 irprog.blocks.remove(block)
-                irprog.st.removeTree(block.label)
+                pruneSymboltable(block.label)
                 numRemoved++
             }
         }
@@ -24,6 +24,22 @@ class IRUnusedCodeRemover(
         return numRemoved
     }
 
+    private fun pruneSymboltable(blockLabel: String) {
+        // we could clean up the SymbolTable as well, but ONLY if these symbols aren't referenced somewhere still in an instruction
+        val prefix = "$blockLabel."
+        val blockVars = irprog.st.allVariables().filter { it.name.startsWith(prefix) }
+        blockVars.forEach { stVar ->
+            irprog. allSubs().flatMap { it.chunks }.forEach { chunk ->
+                chunk.instructions.forEach { ins ->
+                    if(ins.labelSymbol == stVar.name) {
+                        return
+                    }
+                }
+            }
+        }
+        irprog.st.removeTree(blockLabel)
+    }
+
     private fun removeUnusedSubroutines(): Int {
         val allLabeledChunks = mutableMapOf<String, IRCodeChunkBase>()
         irprog.foreachCodeChunk { chunk ->
@@ -34,7 +50,7 @@ class IRUnusedCodeRemover(
         irprog.blocks.forEach { block ->
             block.children.filterIsInstance<IRSubroutine>().reversed().forEach { sub ->
                 if(sub.isEmpty()) {
-                    if(!sub.position.file.startsWith(libraryFilePrefix)) {
+                    if(!sub.position.file.startsWith(LIBRARYFILEPREFIX)) {
                         errors.warn("unused subroutine ${sub.label}", sub.position)
                     }
                     block.children.remove(sub)
@@ -55,7 +71,7 @@ class IRUnusedCodeRemover(
         irprog.blocks.forEach { block ->
             block.children.filterIsInstance<IRAsmSubroutine>().reversed().forEach { sub ->
                 if(sub.isEmpty()) {
-                    if(!sub.position.file.startsWith(libraryFilePrefix)) {
+                    if(!sub.position.file.startsWith(LIBRARYFILEPREFIX)) {
                         errors.warn("unused asmsubroutine ${sub.label}", sub.position)
                     }
                     block.children.remove(sub)
@@ -119,9 +135,22 @@ class IRUnusedCodeRemover(
     }
 
     private fun removeUnreachable(allLabeledChunks: MutableMap<String, IRCodeChunkBase>): Int {
-        val entrypointSub = irprog.blocks.single { it.label=="main" }.children.single { it is IRSubroutine && it.label=="main.start" }
+        val entrypointSub = irprog.blocks.single { it.label=="main" }
+            .children.single { it is IRSubroutine && it.label=="main.start" }
         val reachable = mutableSetOf((entrypointSub as IRSubroutine).chunks.first())
 
+        // all chunks referenced in array initializer values are also 'reachable':
+        irprog.st.allVariables()
+            .filter { !it.uninitialized }
+            .forEach {
+                it.onetimeInitializationArrayValue?.let { array ->
+                    array.forEach {elt ->
+                        if(elt.addressOfSymbol!=null && irprog.st.lookup(elt.addressOfSymbol!!)==null)
+                            reachable.add(irprog.getChunkWithLabel(elt.addressOfSymbol!!))
+                    }
+                }
+            }
+
         fun grow() {
             val new = mutableSetOf<IRCodeChunkBase>()
             reachable.forEach {
@@ -150,6 +179,18 @@ class IRUnusedCodeRemover(
     private fun removeSimpleUnlinked(allLabeledChunks: Map<String, IRCodeChunkBase>): Int {
         val linkedChunks = mutableSetOf<IRCodeChunkBase>()
 
+        // all chunks referenced in array initializer values are linked as well!:
+        irprog.st.allVariables()
+            .filter { !it.uninitialized }
+            .forEach {
+                it.onetimeInitializationArrayValue?.let { array ->
+                    array.forEach {elt ->
+                        if(elt.addressOfSymbol!=null && irprog.st.lookup(elt.addressOfSymbol!!)==null)
+                            linkedChunks += irprog.getChunkWithLabel(elt.addressOfSymbol!!)
+                    }
+                }
+            }
+
         irprog.foreachCodeChunk { chunk ->
             chunk.next?.let { next -> linkedChunks += next }
             chunk.instructions.forEach {

codeGenIntermediate/src/prog8/codegen/vm/VmCodeGen.kt

@@ -16,7 +16,7 @@ class VmCodeGen: ICodeGeneratorBackend {
         symbolTable: SymbolTable,
         options: CompilationOptions,
         errors: IErrorReporter
-    ): IAssemblyProgram? {
+    ): IAssemblyProgram {
         val irCodeGen = IRCodeGen(program, symbolTable, options, errors)
         val irProgram = irCodeGen.generate()
         return VmAssemblyProgram(irProgram.name, irProgram)

codeGenIntermediate/test/Dummies.kt

@@ -43,6 +43,10 @@ internal class ErrorReporterForTests(private val throwExceptionAtReportIfErrors:
             warnings.add(text)
     }
 
+    override fun undefined(symbol: List<String>, position: Position) {
+        err("undefined symbol: ${symbol.joinToString(".")}", position)
+    }
+
     override fun noErrors(): Boolean  = errors.isEmpty()
 
     override fun report() {

codeGenIntermediate/test/TestVmCodeGen.kt

@@ -41,7 +41,7 @@ class TestVmCodeGen: FunSpec({
 //}
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         sub.add(PtVariable("pi", DataType.UBYTE, ZeropageWish.DONTCARE, PtNumber(DataType.UBYTE, 0.0, Position.DUMMY), null, Position.DUMMY))
         sub.add(PtVariable("particleX", DataType.ARRAY_UB, ZeropageWish.DONTCARE, null, 3u, Position.DUMMY))
@@ -91,7 +91,7 @@ class TestVmCodeGen: FunSpec({
         program.add(block)
 
         // define the "cx16.r0" virtual register
-        val cx16block = PtBlock("cx16", null, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val cx16block = PtBlock("cx16", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         cx16block.add(PtMemMapped("r0", DataType.UWORD, 100u, null, Position.DUMMY))
         program.add(cx16block)
 
@@ -120,7 +120,7 @@ class TestVmCodeGen: FunSpec({
 //}
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         sub.add(PtVariable("f1", DataType.FLOAT, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
         val if1 = PtIfElse(Position.DUMMY)
@@ -183,7 +183,7 @@ class TestVmCodeGen: FunSpec({
 //}
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         sub.add(PtVariable("f1", DataType.FLOAT, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
         val if1 = PtIfElse(Position.DUMMY)
@@ -242,7 +242,7 @@ class TestVmCodeGen: FunSpec({
 //}
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         sub.add(PtVariable("f1", DataType.FLOAT, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
         val if1 = PtIfElse(Position.DUMMY)
@@ -289,7 +289,7 @@ class TestVmCodeGen: FunSpec({
 //}
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         sub.add(PtVariable("sb1", DataType.BYTE, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
         val if1 = PtIfElse(Position.DUMMY)
@@ -352,7 +352,7 @@ class TestVmCodeGen: FunSpec({
 //}
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         sub.add(PtVariable("sb1", DataType.BYTE, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
         val if1 = PtIfElse(Position.DUMMY)
@@ -411,7 +411,7 @@ class TestVmCodeGen: FunSpec({
 //}
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         sub.add(PtVariable("ub1", DataType.UBYTE, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
         val if1 = PtIfElse(Position.DUMMY)
@@ -451,7 +451,7 @@ class TestVmCodeGen: FunSpec({
 //}
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val romsub = PtAsmSub("routine", 0x5000u, setOf(CpuRegister.Y), emptyList(), emptyList(), false, Position.DUMMY)
         block.add(romsub)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)

codeOptimizers/src/prog8/optimizer/BinExprSplitter.kt

@@ -1,71 +0,0 @@
-package prog8.optimizer
-
-import prog8.ast.IStatementContainer
-import prog8.ast.Node
-import prog8.ast.Program
-import prog8.ast.expressions.BinaryExpression
-import prog8.ast.statements.AssignTarget
-import prog8.ast.statements.Assignment
-import prog8.ast.statements.AssignmentOrigin
-import prog8.ast.walk.AstWalker
-import prog8.ast.walk.IAstModification
-import prog8.code.core.AugmentAssignmentOperators
-import prog8.code.core.CompilationOptions
-import prog8.code.core.DataType
-import prog8.code.target.VMTarget
-
-
-class BinExprSplitter(private val program: Program, private val options: CompilationOptions) : AstWalker() {
-
-    override fun after(assignment: Assignment, parent: Node): Iterable<IAstModification> {
-
-        if(options.compTarget.name == VMTarget.NAME)
-            return noModifications  // don't split expressions when targeting the vm codegen, it handles nested expressions well
-
-        if(assignment.value.inferType(program) istype DataType.FLOAT && !options.optimizeFloatExpressions)
-            return noModifications
-
-        val binExpr = assignment.value as? BinaryExpression
-        if (binExpr != null) {
-
-            if(binExpr.operator in AugmentAssignmentOperators && isSimpleTarget(assignment.target)) {
-                if(assignment.target isSameAs binExpr.right)
-                    return noModifications
-                if(assignment.target isSameAs binExpr.left) {
-                    if(binExpr.right.isSimple)
-                        return noModifications
-                    val leftBx = binExpr.left as? BinaryExpression
-                    if(leftBx!=null && (!leftBx.left.isSimple || !leftBx.right.isSimple))
-                        return noModifications
-                    val rightBx = binExpr.right as? BinaryExpression
-                    if(rightBx!=null && (!rightBx.left.isSimple || !rightBx.right.isSimple))
-                        return noModifications
-                }
-
-                if(binExpr.right.isSimple) {
-                    val firstAssign = Assignment(assignment.target.copy(), binExpr.left, AssignmentOrigin.OPTIMIZER, binExpr.left.position)
-                    val targetExpr = assignment.target.toExpression()
-                    val augExpr = BinaryExpression(targetExpr, binExpr.operator, binExpr.right, binExpr.right.position)
-                    return listOf(
-                        IAstModification.ReplaceNode(binExpr, augExpr, assignment),
-                        IAstModification.InsertBefore(assignment, firstAssign, assignment.parent as IStatementContainer)
-                    )
-                }
-            }
-
-            // Further unraveling of binary expressions is really complicated here and
-            // often results in much bigger code, thereby defeating the purpose a bit.
-            // All in all this should probably be fixed in a better code generation backend
-            // that doesn't require this at all.
-        }
-
-        return noModifications
-    }
-
-    private fun isSimpleTarget(target: AssignTarget) =
-            if (target.identifier!=null || target.memoryAddress!=null)
-                !target.isIOAddress(options.compTarget.machine)
-            else
-                false
-
-}

codeOptimizers/src/prog8/optimizer/ConstantFoldingOptimizer.kt

@@ -3,16 +3,17 @@ package prog8.optimizer
 import prog8.ast.Node
 import prog8.ast.Program
 import prog8.ast.base.FatalAstException
-import prog8.ast.base.UndefinedSymbolError
 import prog8.ast.expressions.*
 import prog8.ast.maySwapOperandOrder
 import prog8.ast.statements.ForLoop
+import prog8.ast.statements.RepeatLoop
 import prog8.ast.statements.VarDecl
 import prog8.ast.statements.VarDeclType
 import prog8.ast.walk.AstWalker
 import prog8.ast.walk.IAstModification
 import prog8.code.core.AssociativeOperators
 import prog8.code.core.DataType
+import kotlin.math.floor
 
 
 class ConstantFoldingOptimizer(private val program: Program) : AstWalker() {
@@ -234,6 +235,47 @@ class ConstantFoldingOptimizer(private val program: Program) : AstWalker() {
             }
         }
 
+        if(rightconst!=null && (expr.operator=="<<" || expr.operator==">>")) {
+            val dt = expr.left.inferType(program)
+            if(dt.isBytes && rightconst.number>=8) {
+                if(dt.istype(DataType.UBYTE)) {
+                    val zeroUB = NumericLiteral(DataType.UBYTE, 0.0, expr.position)
+                    modifications.add(IAstModification.ReplaceNode(expr, zeroUB, parent))
+                } else {
+                    if(leftconst!=null) {
+                        val zeroB = NumericLiteral(DataType.BYTE, 0.0, expr.position)
+                        val minusoneB = NumericLiteral(DataType.BYTE, -1.0, expr.position)
+                        if(leftconst.number<0.0) {
+                            if(expr.operator=="<<")
+                                modifications.add(IAstModification.ReplaceNode(expr, zeroB, parent))
+                            else
+                                modifications.add(IAstModification.ReplaceNode(expr, minusoneB, parent))
+                        } else {
+                            modifications.add(IAstModification.ReplaceNode(expr, zeroB, parent))
+                        }
+                    }
+                }
+            }
+            else if(dt.isWords && rightconst.number>=16) {
+                if(dt.istype(DataType.UWORD)) {
+                    val zeroUW = NumericLiteral(DataType.UWORD, 0.0, expr.position)
+                    modifications.add(IAstModification.ReplaceNode(expr, zeroUW, parent))
+                } else {
+                    if(leftconst!=null) {
+                        val zeroW = NumericLiteral(DataType.WORD, 0.0, expr.position)
+                        val minusoneW = NumericLiteral(DataType.WORD, -1.0, expr.position)
+                        if(leftconst.number<0.0) {
+                            if(expr.operator=="<<")
+                                modifications.add(IAstModification.ReplaceNode(expr, zeroW, parent))
+                            else
+                                modifications.add(IAstModification.ReplaceNode(expr, minusoneW, parent))
+                        } else {
+                            modifications.add(IAstModification.ReplaceNode(expr, zeroW, parent))
+                        }
+                    }
+                }
+            }
+        }
 
         return modifications
     }
@@ -307,7 +349,7 @@ class ConstantFoldingOptimizer(private val program: Program) : AstWalker() {
         val rangeTo = iterableRange.to as? NumericLiteral
         if(rangeFrom==null || rangeTo==null) return noModifications
 
-        val loopvar = forLoop.loopVar.targetVarDecl(program) ?: throw UndefinedSymbolError(forLoop.loopVar)
+        val loopvar = forLoop.loopVar.targetVarDecl(program) ?: return noModifications
 
         val stepLiteral = iterableRange.step as? NumericLiteral
         when(loopvar.datatype) {
@@ -364,6 +406,16 @@ class ConstantFoldingOptimizer(private val program: Program) : AstWalker() {
         return noModifications
     }
 
+    override fun after(repeatLoop: RepeatLoop, parent: Node): Iterable<IAstModification> {
+        val count = (repeatLoop.iterations as? NumericLiteral)?.number
+        if(count!=null && floor(count)!=count) {
+            val integer = NumericLiteral.optimalInteger(count.toInt(), repeatLoop.position)
+            repeatLoop.iterations = integer
+            integer.linkParents(repeatLoop)
+        }
+        return noModifications
+    }
+
     private class ShuffleOperands(val expr: BinaryExpression,
                                   val exprOperator: String?,
                                   val subExpr: BinaryExpression,
@@ -576,5 +628,4 @@ class ConstantFoldingOptimizer(private val program: Program) : AstWalker() {
             return null
         }
     }
-
 }

codeOptimizers/src/prog8/optimizer/ConstantIdentifierReplacer.kt

@@ -53,7 +53,8 @@ class VarConstantValueTypeAdjuster(private val program: Program, private val err
         }
 
         if(to==null) {
-            if(!range.to.inferType(program).isInteger)
+            val toType = range.to.inferType(program)
+            if(toType.isKnown && !range.to.inferType(program).isInteger)
                 errors.err("range expression to value must be integer", range.to.position)
         } else if(to-to.toInt()>0) {
             errors.err("range expression to value must be integer", range.to.position)
@@ -279,10 +280,16 @@ internal class ConstantIdentifierReplacer(private val program: Program, private
                     val rangeExpr = decl.value as? RangeExpression
                     if(rangeExpr!=null) {
                         // convert the initializer range expression to an actual array
+                        val constRange = rangeExpr.toConstantIntegerRange()
+                        if(constRange?.isEmpty()==true) {
+                            if(constRange.first>constRange.last && constRange.step>=0)
+                                errors.err("descending range with positive step", decl.value?.position!!)
+                            else if(constRange.first<constRange.last && constRange.step<=0)
+                                errors.err("ascending range with negative step", decl.value?.position!!)
+                        }
                         val declArraySize = decl.arraysize?.constIndex()
                         if(declArraySize!=null && declArraySize!=rangeExpr.size())
                             errors.err("range expression size (${rangeExpr.size()}) doesn't match declared array size ($declArraySize)", decl.value?.position!!)
-                        val constRange = rangeExpr.toConstantIntegerRange()
                         if(constRange!=null) {
                             val eltType = rangeExpr.inferType(program).getOr(DataType.UBYTE)
                             val newValue = if(eltType in ByteDatatypes) {

codeOptimizers/src/prog8/optimizer/ExpressionSimplifier.kt

@@ -18,7 +18,7 @@ import kotlin.math.abs
 import kotlin.math.log2
 import kotlin.math.pow
 
-// TODO add more peephole expression optimizations? Investigate what optimizations binaryen has
+// TODO add more peephole expression optimizations? Investigate what optimizations binaryen has?
 
 class ExpressionSimplifier(private val program: Program,
                            private val errors: IErrorReporter,

codeOptimizers/src/prog8/optimizer/Extensions.kt

@@ -43,9 +43,9 @@ fun Program.constantFold(errors: IErrorReporter, compTarget: ICompilationTarget)
 
 fun Program.optimizeStatements(errors: IErrorReporter,
                                functions: IBuiltinFunctions,
-                               compTarget: ICompilationTarget
+                               options: CompilationOptions
 ): Int {
-    val optimizer = StatementOptimizer(this, errors, functions, compTarget)
+    val optimizer = StatementOptimizer(this, errors, functions, options)
     optimizer.visit(this)
     val optimizationCount = optimizer.applyModifications()
 
@@ -54,8 +54,8 @@ fun Program.optimizeStatements(errors: IErrorReporter,
     return optimizationCount
 }
 
-fun Program.inlineSubroutines(): Int {
-    val inliner = Inliner(this)
+fun Program.inlineSubroutines(options: CompilationOptions): Int {
+    val inliner = Inliner(this, options)
     inliner.visit(this)
     return inliner.applyModifications()
 }
@@ -65,9 +65,3 @@ fun Program.simplifyExpressions(errors: IErrorReporter, target: ICompilationTarg
     opti.visit(this)
     return opti.applyModifications()
 }
-
-fun Program.splitBinaryExpressions(options: CompilationOptions) : Int {
-    val opti = BinExprSplitter(this, options)
-    opti.visit(this)
-    return opti.applyModifications()
-}

codeOptimizers/src/prog8/optimizer/Inliner.kt

@@ -8,7 +8,9 @@ import prog8.ast.statements.*
 import prog8.ast.walk.AstWalker
 import prog8.ast.walk.IAstModification
 import prog8.ast.walk.IAstVisitor
+import prog8.code.core.CompilationOptions
 import prog8.code.core.InternalCompilerException
+import prog8.code.target.VMTarget
 
 
 private  fun isEmptyReturn(stmt: Statement): Boolean = stmt is Return && stmt.value==null
@@ -16,7 +18,7 @@ private  fun isEmptyReturn(stmt: Statement): Boolean = stmt is Return && stmt.va
 
 // inliner potentially enables *ONE LINED* subroutines, wihtout to be inlined.
 
-class Inliner(val program: Program): AstWalker() {
+class Inliner(private val program: Program, private val options: CompilationOptions): AstWalker() {
 
     class DetermineInlineSubs(val program: Program): IAstVisitor {
         private val modifications = mutableListOf<IAstModification>()
@@ -211,7 +213,7 @@ class Inliner(val program: Program): AstWalker() {
 
     override fun after(functionCallStatement: FunctionCallStatement, parent: Node): Iterable<IAstModification>  {
         val sub = functionCallStatement.target.targetStatement(program) as? Subroutine
-        return if(sub==null)
+        return if(sub==null || !canInline(sub, functionCallStatement))
             noModifications
         else
             possibleInlineFcallStmt(sub, functionCallStatement, parent)
@@ -219,7 +221,7 @@ class Inliner(val program: Program): AstWalker() {
 
     override fun before(functionCallExpr: FunctionCallExpression, parent: Node): Iterable<IAstModification> {
         val sub = functionCallExpr.target.targetStatement(program) as? Subroutine
-        if(sub!=null && sub.inline && sub.parameters.isEmpty()) {
+        if(sub!=null && sub.inline && sub.parameters.isEmpty() && canInline(sub, functionCallExpr)) {
             require(sub.statements.size == 1 || (sub.statements.size == 2 && isEmptyReturn(sub.statements[1]))) {
                 "invalid inline sub at ${sub.position}"
             }
@@ -246,5 +248,17 @@ class Inliner(val program: Program): AstWalker() {
         return noModifications
     }
 
+    private fun canInline(sub: Subroutine, fcall: IFunctionCall): Boolean {
+        if(!sub.inline)
+            return false
+        if(options.compTarget.name!=VMTarget.NAME) {
+            val stmt = sub.statements.single()
+            if (stmt is IFunctionCall) {
+                val existing = (fcall as Node).definingScope.lookup(stmt.target.nameInSource.take(1))
+                return existing !is VarDecl
+            }
+        }
+        return true
+    }
 }
 

codeOptimizers/src/prog8/optimizer/StatementOptimizer.kt

@@ -13,7 +13,7 @@ import kotlin.math.floor
 class StatementOptimizer(private val program: Program,
                          private val errors: IErrorReporter,
                          private val functions: IBuiltinFunctions,
-                         private val compTarget: ICompilationTarget
+                         private val options: CompilationOptions
 ) : AstWalker() {
 
     override fun after(functionCallStatement: FunctionCallStatement, parent: Node): Iterable<IAstModification> {
@@ -39,7 +39,7 @@ class StatementOptimizer(private val program: Program,
                 if(string!=null) {
                     val pos = functionCallStatement.position
                     if (string.value.length == 1) {
-                        val firstCharEncoded = compTarget.encodeString(string.value, string.encoding)[0]
+                        val firstCharEncoded = options.compTarget.encodeString(string.value, string.encoding)[0]
                         val chrout = FunctionCallStatement(
                             IdentifierReference(listOf("txt", "chrout"), pos),
                             mutableListOf(NumericLiteral(DataType.UBYTE, firstCharEncoded.toDouble(), pos)),
@@ -51,7 +51,7 @@ class StatementOptimizer(private val program: Program,
                             IAstModification.Remove(stringDecl, stringDecl.parent as IStatementContainer)
                         )
                     } else if (string.value.length == 2) {
-                        val firstTwoCharsEncoded = compTarget.encodeString(string.value.take(2), string.encoding)
+                        val firstTwoCharsEncoded = options.compTarget.encodeString(string.value.take(2), string.encoding)
                         val chrout1 = FunctionCallStatement(
                             IdentifierReference(listOf("txt", "chrout"), pos),
                             mutableListOf(NumericLiteral(DataType.UBYTE, firstTwoCharsEncoded[0].toDouble(), pos)),
@@ -108,6 +108,14 @@ class StatementOptimizer(private val program: Program,
             }
         }
 
+        // remove obvious dangling elses (else after a return)
+        if(ifElse.elsepart.isNotEmpty() && ifElse.truepart.statements.singleOrNull() is Return) {
+            val elsePart = AnonymousScope(ifElse.elsepart.statements, ifElse.elsepart.position)
+            return listOf(
+                IAstModification.ReplaceNode(ifElse.elsepart, AnonymousScope(mutableListOf(), ifElse.elsepart.position), ifElse),
+                IAstModification.InsertAfter(ifElse, elsePart, parent as IStatementContainer)
+            )
+        }
         return noModifications
     }
 
@@ -141,7 +149,7 @@ class StatementOptimizer(private val program: Program,
                 val size = sv.value.length
                 if(size==1) {
                     // loop over string of length 1 -> just assign the single character
-                    val character = compTarget.encodeString(sv.value, sv.encoding)[0]
+                    val character = options.compTarget.encodeString(sv.value, sv.encoding)[0]
                     val byte = NumericLiteral(DataType.UBYTE, character.toDouble(), iterable.position)
                     val scope = AnonymousScope(mutableListOf(), forLoop.position)
                     scope.statements.add(Assignment(AssignTarget(forLoop.loopVar, null, null, forLoop.position), byte, AssignmentOrigin.OPTIMIZER, forLoop.position))
@@ -166,16 +174,6 @@ class StatementOptimizer(private val program: Program,
             }
         }
 
-        val iterationCount = forLoop.constIterationCount(program)
-        if(iterationCount!=null) {
-            val loopName = forLoop.loopVar.nameInSource
-            if(!forLoop.iterable.referencesIdentifier(loopName) && !forLoop.body.referencesIdentifier(loopName)) {
-                errors.warn("for loop can be replaced with repeat loop", forLoop.position)
-                val repeat = RepeatLoop(NumericLiteral.optimalNumeric(iterationCount, forLoop.position), forLoop.body, forLoop.position)
-                return listOf(IAstModification.ReplaceNode(forLoop, repeat, parent))
-            }
-        }
-
         return noModifications
     }
 
@@ -326,7 +324,7 @@ class StatementOptimizer(private val program: Program,
                         if (rightCv == 0.0) {
                             return listOf(IAstModification.Remove(assignment, parent as IStatementContainer))
                         } else if (targetDt in IntegerDatatypes && floor(rightCv) == rightCv) {
-                            if (vardeclDt != VarDeclType.MEMORY && rightCv in 1.0..3.0 && compTarget.name!=VMTarget.NAME) {
+                            if (vardeclDt != VarDeclType.MEMORY && rightCv in 1.0..3.0 && options.compTarget.name!=VMTarget.NAME) {
                                 // replace by several INCs if it's not a memory address (inc on a memory mapped register doesn't work very well)
                                 val incs = AnonymousScope(mutableListOf(), assignment.position)
                                 repeat(rightCv.toInt()) {
@@ -340,7 +338,7 @@ class StatementOptimizer(private val program: Program,
                         if (rightCv == 0.0) {
                             return listOf(IAstModification.Remove(assignment, parent as IStatementContainer))
                         } else if (targetDt in IntegerDatatypes && floor(rightCv) == rightCv) {
-                            if (vardeclDt != VarDeclType.MEMORY && rightCv in 1.0..3.0 && compTarget.name!=VMTarget.NAME) {
+                            if (vardeclDt != VarDeclType.MEMORY && rightCv in 1.0..3.0 && options.compTarget.name!=VMTarget.NAME) {
                                 // replace by several DECs if it's not a memory address (dec on a memory mapped register doesn't work very well)
                                 val decs = AnonymousScope(mutableListOf(), assignment.position)
                                 repeat(rightCv.toInt()) {
@@ -386,9 +384,53 @@ class StatementOptimizer(private val program: Program,
     }
 
     override fun before(unrollLoop: UnrollLoop, parent: Node): Iterable<IAstModification> {
-        return if(unrollLoop.iterations<1)
+        val iterations = unrollLoop.iterations.constValue(program)?.number?.toInt()
+        return if(iterations!=null && iterations<1)
             listOf(IAstModification.Remove(unrollLoop, parent as IStatementContainer))
         else
             noModifications
     }
+
+    override fun after(whenStmt: When, parent: Node): Iterable<IAstModification> {
+
+        fun replaceWithIf(condition: Expression, trueBlock: AnonymousScope, elseBlock: AnonymousScope?): List<IAstModification> {
+            val ifStmt = IfElse(condition, trueBlock, elseBlock ?: AnonymousScope(mutableListOf(), whenStmt.position), whenStmt.position)
+            errors.warn("for boolean condition a normal if statement is preferred", whenStmt.position)
+            return listOf(IAstModification.ReplaceNode(whenStmt, ifStmt, parent))
+        }
+
+        if(whenStmt.condition.inferType(program).isBool) {
+            if(whenStmt.choices.all { it.values?.size==1 }) {
+                if (whenStmt.choices.all { it.values!!.single().constValue(program)!!.number in arrayOf(0.0, 1.0) }) {
+                    // it's a when statement on booleans that can just be replaced by an if or if..else.
+                    if (whenStmt.choices.size == 1) {
+                        return if(whenStmt.choices[0].values!![0].constValue(program)!!.number==1.0) {
+                            replaceWithIf(whenStmt.condition, whenStmt.choices[0].statements, null)
+                        } else {
+                            val notCondition = BinaryExpression(whenStmt.condition, "==", NumericLiteral(DataType.UBYTE, 0.0, whenStmt.condition.position), whenStmt.condition.position)
+                            replaceWithIf(notCondition, whenStmt.choices[0].statements, null)
+                        }
+                    } else if (whenStmt.choices.size == 2) {
+                        var trueBlock: AnonymousScope? = null
+                        var elseBlock: AnonymousScope? = null
+                        if(whenStmt.choices[0].values!![0].constValue(program)!!.number==1.0) {
+                            trueBlock = whenStmt.choices[0].statements
+                        } else {
+                            elseBlock = whenStmt.choices[0].statements
+                        }
+                        if(whenStmt.choices[1].values!![0].constValue(program)!!.number==1.0) {
+                            trueBlock = whenStmt.choices[1].statements
+                        } else {
+                            elseBlock = whenStmt.choices[1].statements
+                        }
+                        if(trueBlock!=null && elseBlock!=null) {
+                            return replaceWithIf(whenStmt.condition, trueBlock, elseBlock)
+                        }
+                    }
+                }
+            }
+        }
+        return noModifications
+    }
+
 }

codeOptimizers/src/prog8/optimizer/UnusedCodeRemover.kt

@@ -75,7 +75,9 @@ class UnusedCodeRemover(private val program: Program,
                     if(!block.statements.any { it is Subroutine && it.hasBeenInlined })
                         errors.warn("removing unused block '${block.name}'", block.position)
                 }
-                program.removeInternedStringsFromRemovedBlock(block)
+                if(!block.statements.any { it is Subroutine && it.hasBeenInlined }) {
+                    program.removeInternedStringsFromRemovedBlock(block)
+                }
                 return listOf(IAstModification.Remove(block, parent as IStatementContainer))
             }
         }

compiler/build.gradle

@@ -1,9 +1,10 @@
 plugins {
     id 'java'
     id 'application'
-    id "org.jetbrains.kotlin.jvm"
+    id 'org.jetbrains.kotlin.jvm'
     id 'com.github.johnrengelman.shadow' version '7.1.2'
-    id "io.kotest" version "0.3.9"
+    id 'io.kotest' version '0.3.9'
+    id 'com.peterabeles.gversion' version '1.10.2'
 }
 
 java {
@@ -24,8 +25,6 @@ compileTestKotlin {
     }
 }
 
-def prog8version = rootProject.file('compiler/res/version.txt').text.trim()
-
 dependencies {
     implementation project(':codeCore')
     implementation project(':codeOptimizers')
@@ -38,7 +37,7 @@ dependencies {
     implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8"
     // implementation "org.jetbrains.kotlin:kotlin-reflect"
     implementation 'org.jetbrains.kotlinx:kotlinx-cli:0.3.5'
-    implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.17"
+    implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.18"
 
     testImplementation project(':intermediate')
     testImplementation 'io.kotest:kotest-runner-junit5-jvm:5.6.2'
@@ -82,7 +81,7 @@ application {
 
 shadowJar {
     archiveBaseName = 'prog8compiler'
-    archiveVersion = prog8version
+    archiveVersion = version
     // minimize()
 }
 
@@ -100,4 +99,16 @@ test {
     }
 }
 
+gversion {
+    srcDir       = "src/"           // path is relative to the sub-project by default
+    classPackage = "prog8.buildversion"
+    className    = "BuildVersion"
+    language     = "kotlin"
+}
+
+
 build.finalizedBy installDist, installShadowDist
+
+compileKotlin.dependsOn createVersionFile    // , failDirtyNotSnapshot
+compileJava.dependsOn createVersionFile
+

compiler/res/prog8lib/atari/syslib.p8

@@ -2,6 +2,7 @@
 ; Including memory registers, I/O registers, Basic and Kernal subroutines.
 
 atari {
+    %option no_symbol_prefixing
 
         &uword  NMI_VEC         = $FFFA     ; 6502 nmi vector, determined by the kernal if banked in
         &uword  RESET_VEC       = $FFFC     ; 6502 reset vector, determined by the kernal if banked in
@@ -9,10 +10,11 @@ atari {
 
 }
 
-
 sys {
     ; ------- lowlevel system routines --------
 
+    %option no_symbol_prefixing
+
     const ubyte target = 8         ;  compilation target specifier.  64 = C64, 128 = C128,  16 = CommanderX16, 8 = atari800XL
 
     asmsub  init_system()  {
@@ -216,6 +218,7 @@ _longcopy
 }
 
 cx16 {
+    %option no_symbol_prefixing
 
     ; the sixteen virtual 16-bit registers that the CX16 has defined in the zeropage
     ; they are simulated on the Atari as well but their location in memory is different
@@ -321,4 +324,31 @@ cx16 {
     &byte r13sH = $1b1b
     &byte r14sH = $1b1d
     &byte r15sH = $1b1f
+
+    asmsub save_virtual_registers() clobbers(A,Y) {
+        %asm {{
+            ldy  #31
+    -       lda  cx16.r0,y
+            sta  _cx16_vreg_storage,y
+            dey
+            bpl  -
+            rts
+
+    _cx16_vreg_storage
+            .word 0,0,0,0,0,0,0,0
+            .word 0,0,0,0,0,0,0,0
+        }}
+    }
+
+    asmsub restore_virtual_registers() clobbers(A,Y) {
+        %asm {{
+            ldy  #31
+    -       lda  save_virtual_registers._cx16_vreg_storage,y
+            sta  cx16.r0,y
+            dey
+            bpl  -
+            rts
+        }}
+    }
+
 }

compiler/res/prog8lib/atari/textio.p8

@@ -6,6 +6,8 @@
 
 txt {
 
+    %option no_symbol_prefixing
+
 const ubyte DEFAULT_WIDTH = 40
 const ubyte DEFAULT_HEIGHT = 24
 
@@ -30,7 +32,7 @@ asmsub column(ubyte col @A) clobbers(A, X, Y) {
     }}
 }
 
-asmsub  fill_screen (ubyte char @ A, ubyte color @ Y) clobbers(A)  {
+asmsub  fill_screen (ubyte character @ A, ubyte color @ Y) clobbers(A)  {
     ; ---- fill the character screen with the given fill character and character color.
     ;      (assumes screen and color matrix are at their default addresses)
     ;      TODO
@@ -40,7 +42,7 @@ asmsub  fill_screen (ubyte char @ A, ubyte color @ Y) clobbers(A)  {
     }}
 }
 
-asmsub  clear_screenchars (ubyte char @ A) clobbers(Y)  {
+asmsub  clear_screenchars (ubyte character @ A) clobbers(Y)  {
 	; ---- clear the character screen with the given fill character (leaves colors)
 	;      (assumes screen matrix is at the default address)
 	; TODO
@@ -112,10 +114,10 @@ asmsub  scroll_down  (bool alsocolors @ Pc) clobbers(A)  {
 }
 
 
-romsub $F2B0 = outchar(ubyte char @ A)
+romsub $F2B0 = outchar(ubyte character @ A)
 romsub $F2Fd = waitkey()
 
-asmsub chrout(ubyte char @ A) {
+asmsub chrout(ubyte character @ A) {
 	%asm {{
 		sta  _tmp_outchar+1
 		txa
@@ -151,10 +153,9 @@ asmsub  print (str text @ AY) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  print_ub0  (ubyte value @ A) clobbers(A,Y)  {
+asmsub  print_ub0  (ubyte value @ A) clobbers(A,X,Y)  {
 	; ---- print the ubyte in A in decimal form, with left padding 0s (3 positions total)
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		jsr  conv.ubyte2decimal
 		pha
 		tya
@@ -162,16 +163,13 @@ asmsub  print_ub0  (ubyte value @ A) clobbers(A,Y)  {
 		pla
 		jsr  chrout
 		txa
-		jsr  chrout
-		ldx  P8ZP_SCRATCH_REG
-		rts
+		jmp  chrout
 	}}
 }
 
-asmsub  print_ub  (ubyte value @ A) clobbers(A,Y)  {
+asmsub  print_ub  (ubyte value @ A) clobbers(A,X,Y)  {
 	; ---- print the ubyte in A in decimal form, without left padding 0s
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		jsr  conv.ubyte2decimal
 _print_byte_digits
 		pha
@@ -187,16 +185,13 @@ _print_byte_digits
         beq  _ones
         jsr  chrout
 _ones   txa
-		jsr  chrout
-		ldx  P8ZP_SCRATCH_REG
-		rts
+		jmp  chrout
 	}}
 }
 
-asmsub  print_b  (byte value @ A) clobbers(A,Y)  {
+asmsub  print_b  (byte value @ A) clobbers(A,X,Y)  {
 	; ---- print the byte in A in decimal form, without left padding 0s
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		pha
 		cmp  #0
 		bpl  +
@@ -208,10 +203,9 @@ asmsub  print_b  (byte value @ A) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  print_ubhex  (ubyte value @ A, bool prefix @ Pc) clobbers(A,Y)  {
+asmsub  print_ubhex  (ubyte value @ A, bool prefix @ Pc) clobbers(A,X,Y)  {
 	; ---- print the ubyte in A in hex form (if Carry is set, a radix prefix '$' is printed as well)
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		bcc  +
 		pha
 		lda  #'$'
@@ -220,16 +214,13 @@ asmsub  print_ubhex  (ubyte value @ A, bool prefix @ Pc) clobbers(A,Y)  {
 +		jsr  conv.ubyte2hex
 		jsr  chrout
 		tya
-		jsr  chrout
-		ldx  P8ZP_SCRATCH_REG
-		rts
+		jmp  chrout
 	}}
 }
 
-asmsub  print_ubbin  (ubyte value @ A, bool prefix @ Pc) clobbers(A,Y)  {
+asmsub  print_ubbin  (ubyte value @ A, bool prefix @ Pc) clobbers(A,X,Y)  {
 	; ---- print the ubyte in A in binary form (if Carry is set, a radix prefix '%' is printed as well)
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		sta  P8ZP_SCRATCH_B1
 		bcc  +
 		lda  #'%'
@@ -242,12 +233,11 @@ asmsub  print_ubbin  (ubyte value @ A, bool prefix @ Pc) clobbers(A,Y)  {
 +		jsr  chrout
 		dey
 		bne  -
-		ldx  P8ZP_SCRATCH_REG
 		rts
 	}}
 }
 
-asmsub  print_uwbin  (uword value @ AY, bool prefix @ Pc) clobbers(A,Y)  {
+asmsub  print_uwbin  (uword value @ AY, bool prefix @ Pc) clobbers(A,X,Y)  {
 	; ---- print the uword in A/Y in binary form (if Carry is set, a radix prefix '%' is printed as well)
 	%asm {{
 		pha
@@ -259,7 +249,7 @@ asmsub  print_uwbin  (uword value @ AY, bool prefix @ Pc) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  print_uwhex  (uword value @ AY, bool prefix @ Pc) clobbers(A,Y)  {
+asmsub  print_uwhex  (uword value @ AY, bool prefix @ Pc) clobbers(A,X,Y)  {
 	; ---- print the uword in A/Y in hexadecimal form (4 digits)
 	;      (if Carry is set, a radix prefix '$' is printed as well)
 	%asm {{
@@ -272,10 +262,9 @@ asmsub  print_uwhex  (uword value @ AY, bool prefix @ Pc) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  print_uw0  (uword value @ AY) clobbers(A,Y)  {
+asmsub  print_uw0  (uword value @ AY) clobbers(A,X,Y)  {
 	; ---- print the uword in A/Y in decimal form, with left padding 0s (5 positions total)
 	%asm {{
-	    stx  P8ZP_SCRATCH_REG
 		jsr  conv.uword2decimal
 		ldy  #0
 -		lda  conv.uword2decimal.decTenThousands,y
@@ -283,17 +272,14 @@ asmsub  print_uw0  (uword value @ AY) clobbers(A,Y)  {
 		jsr  chrout
 		iny
 		bne  -
-+		ldx  P8ZP_SCRATCH_REG
-		rts
++		rts
 	}}
 }
 
-asmsub  print_uw  (uword value @ AY) clobbers(A,Y)  {
+asmsub  print_uw  (uword value @ AY) clobbers(A,X,Y)  {
 	; ---- print the uword in A/Y in decimal form, without left padding 0s
 	%asm {{
-	    stx  P8ZP_SCRATCH_REG
 		jsr  conv.uword2decimal
-		ldx  P8ZP_SCRATCH_REG
 		ldy  #0
 -		lda  conv.uword2decimal.decTenThousands,y
 		beq  _allzero
@@ -314,7 +300,7 @@ _allzero
 	}}
 }
 
-asmsub  print_w  (word value @ AY) clobbers(A,Y)  {
+asmsub  print_w  (word value @ AY) clobbers(A,X,Y)  {
 	; ---- print the (signed) word in A/Y in decimal form, without left padding 0's
 	%asm {{
 		cpy  #0
@@ -378,7 +364,7 @@ asmsub  getclr  (ubyte col @A, ubyte row @Y) clobbers(Y) -> ubyte @ A {
         }}
 }
 
-sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
+sub  setcc  (ubyte col, ubyte row, ubyte char, ubyte charcolor)  {
 	; ---- set char+color at the given position on the screen
 	; TODO
 	%asm {{
@@ -386,7 +372,7 @@ sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
         }}
 }
 
-asmsub  plot  (ubyte col @ Y, ubyte row @ A) clobbers(A) {
+asmsub  plot  (ubyte col @ Y, ubyte row @ A) {
 	; ---- set cursor at specific position
 	; TODO
 	%asm  {{

compiler/res/prog8lib/c128/graphics.p8

@@ -6,6 +6,8 @@
 ; TODO c128 actually implement the graphics routines. Ideally a way to 'borrow' the code form the C64 version without just copy-pasting that here?
 
 graphics {
+    %option no_symbol_prefixing
+
     const uword WIDTH = 320
     const ubyte HEIGHT = 200
 

compiler/res/prog8lib/c128/syslib.p8

@@ -3,6 +3,7 @@
 
 cbm {
     ; Commodore (CBM) common variables, vectors and kernal routines
+    %option no_symbol_prefixing
 
         &ubyte  TIME_HI         = $a0       ; software jiffy clock, hi byte
         &ubyte  TIME_MID        = $a1       ;  .. mid byte
@@ -86,7 +87,7 @@ romsub $FFC6 = CHKIN(ubyte logical @ X) clobbers(A,X) -> bool @Pc    ; (via 798
 romsub $FFC9 = CHKOUT(ubyte logical @ X) clobbers(A,X)          ; (via 800 ($320)) define an output channel
 romsub $FFCC = CLRCHN() clobbers(A,X)                           ; (via 802 ($322)) restore default devices
 romsub $FFCF = CHRIN() clobbers(X, Y) -> ubyte @ A   ; (via 804 ($324)) input a character (for keyboard, read a whole line from the screen) A=byte read.
-romsub $FFD2 = CHROUT(ubyte char @ A)                           ; (via 806 ($326)) output a character
+romsub $FFD2 = CHROUT(ubyte character @ A)                           ; (via 806 ($326)) output a character
 romsub $FFD5 = LOAD(ubyte verify @ A, uword address @ XY) -> bool @Pc, ubyte @ A, uword @ XY     ; (via 816 ($330)) load from device
 romsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) -> bool @ Pc, ubyte @ A          ; (via 818 ($332)) save to a device
 romsub $FFDB = SETTIM(ubyte low @ A, ubyte middle @ X, ubyte high @ Y)      ; set the software clock
@@ -103,34 +104,26 @@ romsub $FFF3 = IOBASE() -> uword @ XY                           ; read base addr
 
 ; ---- utilities -----
 
-asmsub STOP2() -> ubyte @A  {
+asmsub STOP2() clobbers(X) -> ubyte @A  {
     ; -- check if STOP key was pressed, returns true if so.  More convenient to use than STOP() because that only sets the carry status flag.
     %asm {{
-        txa
-        pha
         jsr  cbm.STOP
         beq  +
-        pla
-        tax
         lda  #0
         rts
-+       pla
-        tax
-        lda  #1
++       lda  #1
         rts
     }}
 }
 
-asmsub RDTIM16() -> uword @AY {
+asmsub RDTIM16() clobbers(X) -> uword @AY {
     ; --  like RDTIM() but only returning the lower 16 bits in AY for convenience
     %asm {{
-        stx  P8ZP_SCRATCH_REG
         jsr  cbm.RDTIM
         pha
         txa
         tay
         pla
-        ldx  P8ZP_SCRATCH_REG
         rts
     }}
 }
@@ -140,6 +133,7 @@ asmsub RDTIM16() -> uword @AY {
 
 c64 {
         ; C64 I/O registers (VIC, SID, CIA)
+    %option no_symbol_prefixing
 
         ; the default locations of the 8 sprite pointers (store address of sprite / 64)
         &ubyte  SPRPTR0         = 2040
@@ -294,6 +288,7 @@ c64 {
 
 c128 {
 ; ---- C128 specific registers ----
+    %option no_symbol_prefixing
 
     &ubyte  VM1     = $0A2C         ; shadow for VUC $d018 in text mode
     &ubyte  VM2     = $0A2D         ; shadow for VIC $d018 in bitmap screen mode
@@ -330,6 +325,7 @@ asmsub  disable_basic() clobbers(A) {
 
 sys {
     ; ------- lowlevel system routines --------
+    %option no_symbol_prefixing
 
     const ubyte target = 128         ;  compilation target specifier.  64 = C64, 128 = C128,  16 = CommanderX16.
 
@@ -339,7 +335,7 @@ asmsub  init_system()  {
     ; This means that the BASIC, KERNAL and CHARGEN ROMs are banked in,
     ; the VIC, SID and CIA chips are reset, screen is cleared, and the default IRQ is set.
     ; Also a different color scheme is chosen to identify ourselves a little.
-    ; Uppercase charset is activated, and all three registers set to 0, status flags cleared.
+    ; Uppercase charset is activated.
     %asm {{
         sei
         cld
@@ -442,10 +438,6 @@ _irq_handler_init
 		sta  IRQ_SCRATCH_ZPWORD2
 		lda  P8ZP_SCRATCH_W2+1
 		sta  IRQ_SCRATCH_ZPWORD2+1
-		; Set X to the bottom 32 bytes of the evaluation stack, to HOPEFULLY not clobber it.
-		; This leaves 128-32=96 stack entries for the main program, and 32 stack entries for the IRQ handler.
-		; We assume IRQ handlers don't contain complex expressions taking up more than that.
-		ldx  #32
 		cld
 		rts
 
@@ -765,113 +757,137 @@ _longcopy
 }
 
 cx16 {
+    %option no_symbol_prefixing
 
     ; the sixteen virtual 16-bit registers that the CX16 has defined in the zeropage
     ; they are simulated on the C128 as well but their location in memory is different
     ; (because there's no room for them in the zeropage)
-    ; $1300-$1bff is unused RAM on C128. We'll use $1a00-$1bff as the lo/hi evalstack.
-    ; the virtual registers are allocated at the bottom of the eval-stack (should be ample space unless
-    ; you're doing insane nesting of expressions...)
-    ; NOTE: the memory location of these registers can change based on the "-esa" compiler option
-    &uword r0  = $1b00
-    &uword r1  = $1b02
-    &uword r2  = $1b04
-    &uword r3  = $1b06
-    &uword r4  = $1b08
-    &uword r5  = $1b0a
-    &uword r6  = $1b0c
-    &uword r7  = $1b0e
-    &uword r8  = $1b10
-    &uword r9  = $1b12
-    &uword r10 = $1b14
-    &uword r11 = $1b16
-    &uword r12 = $1b18
-    &uword r13 = $1b1a
-    &uword r14 = $1b1c
-    &uword r15 = $1b1e
+    ; $1300-$1bff is unused RAM on C128.
+    &uword r0  = $1be0
+    &uword r1  = $1be2
+    &uword r2  = $1be4
+    &uword r3  = $1be6
+    &uword r4  = $1be8
+    &uword r5  = $1bea
+    &uword r6  = $1bec
+    &uword r7  = $1bee
+    &uword r8  = $1bf0
+    &uword r9  = $1bf2
+    &uword r10 = $1bf4
+    &uword r11 = $1bf6
+    &uword r12 = $1bf8
+    &uword r13 = $1bfa
+    &uword r14 = $1bfc
+    &uword r15 = $1bfe
 
-    &word r0s  = $1b00
-    &word r1s  = $1b02
-    &word r2s  = $1b04
-    &word r3s  = $1b06
-    &word r4s  = $1b08
-    &word r5s  = $1b0a
-    &word r6s  = $1b0c
-    &word r7s  = $1b0e
-    &word r8s  = $1b10
-    &word r9s  = $1b12
-    &word r10s = $1b14
-    &word r11s = $1b16
-    &word r12s = $1b18
-    &word r13s = $1b1a
-    &word r14s = $1b1c
-    &word r15s = $1b1e
+    &word r0s  = $1be0
+    &word r1s  = $1be2
+    &word r2s  = $1be4
+    &word r3s  = $1be6
+    &word r4s  = $1be8
+    &word r5s  = $1bea
+    &word r6s  = $1bec
+    &word r7s  = $1bee
+    &word r8s  = $1bf0
+    &word r9s  = $1bf2
+    &word r10s = $1bf4
+    &word r11s = $1bf6
+    &word r12s = $1bf8
+    &word r13s = $1bfa
+    &word r14s = $1bfc
+    &word r15s = $1bfe
 
-    &ubyte r0L  = $1b00
-    &ubyte r1L  = $1b02
-    &ubyte r2L  = $1b04
-    &ubyte r3L  = $1b06
-    &ubyte r4L  = $1b08
-    &ubyte r5L  = $1b0a
-    &ubyte r6L  = $1b0c
-    &ubyte r7L  = $1b0e
-    &ubyte r8L  = $1b10
-    &ubyte r9L  = $1b12
-    &ubyte r10L = $1b14
-    &ubyte r11L = $1b16
-    &ubyte r12L = $1b18
-    &ubyte r13L = $1b1a
-    &ubyte r14L = $1b1c
-    &ubyte r15L = $1b1e
+    &ubyte r0L  = $1be0
+    &ubyte r1L  = $1be2
+    &ubyte r2L  = $1be4
+    &ubyte r3L  = $1be6
+    &ubyte r4L  = $1be8
+    &ubyte r5L  = $1bea
+    &ubyte r6L  = $1bec
+    &ubyte r7L  = $1bee
+    &ubyte r8L  = $1bf0
+    &ubyte r9L  = $1bf2
+    &ubyte r10L = $1bf4
+    &ubyte r11L = $1bf6
+    &ubyte r12L = $1bf8
+    &ubyte r13L = $1bfa
+    &ubyte r14L = $1bfc
+    &ubyte r15L = $1bfe
 
-    &ubyte r0H  = $1b01
-    &ubyte r1H  = $1b03
-    &ubyte r2H  = $1b05
-    &ubyte r3H  = $1b07
-    &ubyte r4H  = $1b09
-    &ubyte r5H  = $1b0b
-    &ubyte r6H  = $1b0d
-    &ubyte r7H  = $1b0f
-    &ubyte r8H  = $1b11
-    &ubyte r9H  = $1b13
-    &ubyte r10H = $1b15
-    &ubyte r11H = $1b17
-    &ubyte r12H = $1b19
-    &ubyte r13H = $1b1b
-    &ubyte r14H = $1b1d
-    &ubyte r15H = $1b1f
+    &ubyte r0H  = $1be1
+    &ubyte r1H  = $1be3
+    &ubyte r2H  = $1be5
+    &ubyte r3H  = $1be7
+    &ubyte r4H  = $1be9
+    &ubyte r5H  = $1beb
+    &ubyte r6H  = $1bed
+    &ubyte r7H  = $1bef
+    &ubyte r8H  = $1bf1
+    &ubyte r9H  = $1bf3
+    &ubyte r10H = $1bf5
+    &ubyte r11H = $1bf7
+    &ubyte r12H = $1bf9
+    &ubyte r13H = $1bfb
+    &ubyte r14H = $1bfd
+    &ubyte r15H = $1bff
 
-    &byte r0sL  = $1b00
-    &byte r1sL  = $1b02
-    &byte r2sL  = $1b04
-    &byte r3sL  = $1b06
-    &byte r4sL  = $1b08
-    &byte r5sL  = $1b0a
-    &byte r6sL  = $1b0c
-    &byte r7sL  = $1b0e
-    &byte r8sL  = $1b10
-    &byte r9sL  = $1b12
-    &byte r10sL = $1b14
-    &byte r11sL = $1b16
-    &byte r12sL = $1b18
-    &byte r13sL = $1b1a
-    &byte r14sL = $1b1c
-    &byte r15sL = $1b1e
+    &byte r0sL  = $1be0
+    &byte r1sL  = $1be2
+    &byte r2sL  = $1be4
+    &byte r3sL  = $1be6
+    &byte r4sL  = $1be8
+    &byte r5sL  = $1bea
+    &byte r6sL  = $1bec
+    &byte r7sL  = $1bee
+    &byte r8sL  = $1bf0
+    &byte r9sL  = $1bf2
+    &byte r10sL = $1bf4
+    &byte r11sL = $1bf6
+    &byte r12sL = $1bf8
+    &byte r13sL = $1bfa
+    &byte r14sL = $1bfc
+    &byte r15sL = $1bfe
 
-    &byte r0sH  = $1b01
-    &byte r1sH  = $1b03
-    &byte r2sH  = $1b05
-    &byte r3sH  = $1b07
-    &byte r4sH  = $1b09
-    &byte r5sH  = $1b0b
-    &byte r6sH  = $1b0d
-    &byte r7sH  = $1b0f
-    &byte r8sH  = $1b11
-    &byte r9sH  = $1b13
-    &byte r10sH = $1b15
-    &byte r11sH = $1b17
-    &byte r12sH = $1b19
-    &byte r13sH = $1b1b
-    &byte r14sH = $1b1d
-    &byte r15sH = $1b1f
+    &byte r0sH  = $1be1
+    &byte r1sH  = $1be3
+    &byte r2sH  = $1be5
+    &byte r3sH  = $1be7
+    &byte r4sH  = $1be9
+    &byte r5sH  = $1beb
+    &byte r6sH  = $1bed
+    &byte r7sH  = $1bef
+    &byte r8sH  = $1bf1
+    &byte r9sH  = $1bf3
+    &byte r10sH = $1bf5
+    &byte r11sH = $1bf7
+    &byte r12sH = $1bf9
+    &byte r13sH = $1bfb
+    &byte r14sH = $1bfd
+    &byte r15sH = $1bff
+
+    asmsub save_virtual_registers() clobbers(A,Y) {
+        %asm {{
+            ldy  #31
+    -       lda  cx16.r0,y
+            sta  _cx16_vreg_storage,y
+            dey
+            bpl  -
+            rts
+
+    _cx16_vreg_storage
+            .word 0,0,0,0,0,0,0,0
+            .word 0,0,0,0,0,0,0,0
+        }}
+    }
+
+    asmsub restore_virtual_registers() clobbers(A,Y) {
+        %asm {{
+            ldy  #31
+    -       lda  save_virtual_registers._cx16_vreg_storage,y
+            sta  cx16.r0,y
+            dey
+            bpl  -
+            rts
+        }}
+    }
 }

compiler/res/prog8lib/c128/textio.p8

@@ -1,10 +1,11 @@
-; Prog8 definitions for the Text I/O and Screen routines for the Commodore-64
+; Prog8 definitions for the Text I/O and Screen routines for the Commodore-128
 
 %import syslib
 %import conv
 
 
 txt {
+    %option no_symbol_prefixing
 
 const ubyte DEFAULT_WIDTH = 40
 const ubyte DEFAULT_HEIGHT = 25
@@ -96,13 +97,12 @@ sub uppercase() {
     c128.VM1 &= ~2
 }
 
-asmsub  scroll_left  (bool alsocolors @ Pc) clobbers(A, Y)  {
+asmsub  scroll_left  (bool alsocolors @ Pc) clobbers(A, X, Y)  {
 	; ---- scroll the whole screen 1 character to the left
 	;      contents of the rightmost column are unchanged, you should clear/refill this yourself
 	;      Carry flag determines if screen color data must be scrolled too
 
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		bcc _scroll_screen
 
 +               ; scroll the screen and the color memory
@@ -132,17 +132,15 @@ _scroll_screen  ; scroll only the screen memory
 		dey
 		bpl  -
 
-		ldx  P8ZP_SCRATCH_REG
 		rts
 	}}
 }
 
-asmsub  scroll_right  (bool alsocolors @ Pc) clobbers(A)  {
+asmsub  scroll_right  (bool alsocolors @ Pc) clobbers(A,X)  {
 	; ---- scroll the whole screen 1 character to the right
 	;      contents of the leftmost column are unchanged, you should clear/refill this yourself
 	;      Carry flag determines if screen color data must be scrolled too
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		bcc  _scroll_screen
 
 +               ; scroll the screen and the color memory
@@ -168,17 +166,15 @@ _scroll_screen  ; scroll only the screen memory
 		dex
 		bpl  -
 
-		ldx  P8ZP_SCRATCH_REG
 		rts
 	}}
 }
 
-asmsub  scroll_up  (bool alsocolors @ Pc) clobbers(A)  {
+asmsub  scroll_up  (bool alsocolors @ Pc) clobbers(A,X)  {
 	; ---- scroll the whole screen 1 character up
 	;      contents of the bottom row are unchanged, you should refill/clear this yourself
 	;      Carry flag determines if screen color data must be scrolled too
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		bcc  _scroll_screen
 
 +               ; scroll the screen and the color memory
@@ -204,17 +200,15 @@ _scroll_screen  ; scroll only the screen memory
 		dex
 		bpl  -
 
-		ldx  P8ZP_SCRATCH_REG
 		rts
 	}}
 }
 
-asmsub  scroll_down  (bool alsocolors @ Pc) clobbers(A)  {
+asmsub  scroll_down  (bool alsocolors @ Pc) clobbers(A,X)  {
 	; ---- scroll the whole screen 1 character down
 	;      contents of the top row are unchanged, you should refill/clear this yourself
 	;      Carry flag determines if screen color data must be scrolled too
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		bcc  _scroll_screen
 
 +               ; scroll the screen and the color memory
@@ -240,7 +234,6 @@ _scroll_screen  ; scroll only the screen memory
 		dex
 		bpl  -
 
-		ldx  P8ZP_SCRATCH_REG
 		rts
 	}}
 }
@@ -265,10 +258,9 @@ asmsub  print (str text @ AY) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  print_ub0  (ubyte value @ A) clobbers(A,Y)  {
+asmsub  print_ub0  (ubyte value @ A) clobbers(A,X,Y)  {
 	; ---- print the ubyte in A in decimal form, with left padding 0s (3 positions total)
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		jsr  conv.ubyte2decimal
 		pha
 		tya
@@ -276,16 +268,13 @@ asmsub  print_ub0  (ubyte value @ A) clobbers(A,Y)  {
 		pla
 		jsr  cbm.CHROUT
 		txa
-		jsr  cbm.CHROUT
-		ldx  P8ZP_SCRATCH_REG
-		rts
+		jmp  cbm.CHROUT
 	}}
 }
 
-asmsub  print_ub  (ubyte value @ A) clobbers(A,Y)  {
+asmsub  print_ub  (ubyte value @ A) clobbers(A,X,Y)  {
 	; ---- print the ubyte in A in decimal form, without left padding 0s
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		jsr  conv.ubyte2decimal
 _print_byte_digits
 		pha
@@ -301,16 +290,13 @@ _print_byte_digits
         beq  _ones
         jsr  cbm.CHROUT
 _ones   txa
-		jsr  cbm.CHROUT
-		ldx  P8ZP_SCRATCH_REG
-		rts
+		jmp  cbm.CHROUT
 	}}
 }
 
-asmsub  print_b  (byte value @ A) clobbers(A,Y)  {
+asmsub  print_b  (byte value @ A) clobbers(A,X,Y)  {
 	; ---- print the byte in A in decimal form, without left padding 0s
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		pha
 		cmp  #0
 		bpl  +
@@ -322,10 +308,9 @@ asmsub  print_b  (byte value @ A) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  print_ubhex  (ubyte value @ A, bool prefix @ Pc) clobbers(A,Y)  {
+asmsub  print_ubhex  (ubyte value @ A, bool prefix @ Pc) clobbers(A,X,Y)  {
 	; ---- print the ubyte in A in hex form (if Carry is set, a radix prefix '$' is printed as well)
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		bcc  +
 		pha
 		lda  #'$'
@@ -334,16 +319,13 @@ asmsub  print_ubhex  (ubyte value @ A, bool prefix @ Pc) clobbers(A,Y)  {
 +		jsr  conv.ubyte2hex
 		jsr  cbm.CHROUT
 		tya
-		jsr  cbm.CHROUT
-		ldx  P8ZP_SCRATCH_REG
-		rts
+		jmp  cbm.CHROUT
 	}}
 }
 
-asmsub  print_ubbin  (ubyte value @ A, bool prefix @ Pc) clobbers(A,Y)  {
+asmsub  print_ubbin  (ubyte value @ A, bool prefix @ Pc) clobbers(A,X,Y)  {
 	; ---- print the ubyte in A in binary form (if Carry is set, a radix prefix '%' is printed as well)
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		sta  P8ZP_SCRATCH_B1
 		bcc  +
 		lda  #'%'
@@ -356,12 +338,11 @@ asmsub  print_ubbin  (ubyte value @ A, bool prefix @ Pc) clobbers(A,Y)  {
 +		jsr  cbm.CHROUT
 		dey
 		bne  -
-		ldx  P8ZP_SCRATCH_REG
 		rts
 	}}
 }
 
-asmsub  print_uwbin  (uword value @ AY, bool prefix @ Pc) clobbers(A,Y)  {
+asmsub  print_uwbin  (uword value @ AY, bool prefix @ Pc) clobbers(A,X,Y)  {
 	; ---- print the uword in A/Y in binary form (if Carry is set, a radix prefix '%' is printed as well)
 	%asm {{
 		pha
@@ -373,7 +354,7 @@ asmsub  print_uwbin  (uword value @ AY, bool prefix @ Pc) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  print_uwhex  (uword value @ AY, bool prefix @ Pc) clobbers(A,Y)  {
+asmsub  print_uwhex  (uword value @ AY, bool prefix @ Pc) clobbers(A,X,Y)  {
 	; ---- print the uword in A/Y in hexadecimal form (4 digits)
 	;      (if Carry is set, a radix prefix '$' is printed as well)
 	%asm {{
@@ -386,10 +367,9 @@ asmsub  print_uwhex  (uword value @ AY, bool prefix @ Pc) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  print_uw0  (uword value @ AY) clobbers(A,Y)  {
+asmsub  print_uw0  (uword value @ AY) clobbers(A,X,Y)  {
 	; ---- print the uword in A/Y in decimal form, with left padding 0s (5 positions total)
 	%asm {{
-	    stx  P8ZP_SCRATCH_REG
 		jsr  conv.uword2decimal
 		ldy  #0
 -		lda  conv.uword2decimal.decTenThousands,y
@@ -397,17 +377,14 @@ asmsub  print_uw0  (uword value @ AY) clobbers(A,Y)  {
 		jsr  cbm.CHROUT
 		iny
 		bne  -
-+		ldx  P8ZP_SCRATCH_REG
-		rts
++		rts
 	}}
 }
 
-asmsub  print_uw  (uword value @ AY) clobbers(A,Y)  {
+asmsub  print_uw  (uword value @ AY) clobbers(A,X,Y)  {
 	; ---- print the uword in A/Y in decimal form, without left padding 0s
 	%asm {{
-	    stx  P8ZP_SCRATCH_REG
 		jsr  conv.uword2decimal
-		ldx  P8ZP_SCRATCH_REG
 		ldy  #0
 -		lda  conv.uword2decimal.decTenThousands,y
 		beq  _allzero
@@ -428,7 +405,7 @@ _allzero
 	}}
 }
 
-asmsub  print_w  (word value @ AY) clobbers(A,Y)  {
+asmsub  print_w  (word value @ AY) clobbers(A,X,Y)  {
 	; ---- print the (signed) word in A/Y in decimal form, without left padding 0's
 	%asm {{
 		cpy  #0
@@ -489,7 +466,7 @@ asmsub  setchr  (ubyte col @X, ubyte row @Y, ubyte character @A) clobbers(A, Y)
 _mod		sta  $ffff		; modified
 		rts
 
-_screenrows	.word  $0400 + range(0, 1000, 40)
+_screenrows	.word  cbm.Screen + range(0, 1000, 40)
 	}}
 }
 
@@ -556,7 +533,7 @@ _mod		lda  $ffff		; modified
 	}}
 }
 
-sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
+sub  setcc  (ubyte col, ubyte row, ubyte character, ubyte charcolor)  {
 	; ---- set char+color at the given position on the screen
 	%asm {{
 		lda  row
@@ -568,13 +545,13 @@ sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
 		sta  _colormod+2
 		lda  setchr._screenrows,y
 		clc
-		adc  column
+		adc  col
 		sta  _charmod+1
 		sta  _colormod+1
 		bcc  +
 		inc  _charmod+2
 		inc  _colormod+2
-+		lda  char
++		lda  character
 _charmod	sta  $ffff		; modified
 		lda  charcolor
 _colormod	sta  $ffff		; modified
@@ -582,15 +559,10 @@ _colormod	sta  $ffff		; modified
 	}}
 }
 
-asmsub  plot  (ubyte col @ Y, ubyte row @ A) clobbers(A) {
-	; ---- safe wrapper around PLOT kernal routine, to save the X register.
+asmsub  plot  (ubyte col @ Y, ubyte row @ X) {
 	%asm  {{
-		stx  P8ZP_SCRATCH_REG
-		tax
 		clc
-		jsr  cbm.PLOT
-		ldx  P8ZP_SCRATCH_REG
-		rts
+		jmp  cbm.PLOT
 	}}
 }
 

compiler/res/prog8lib/c64/floats.asm

@@ -5,13 +5,11 @@ FL_ZERO_const	.byte  0,0,0,0,0		; 0.0
 FL_LOG2_const	.byte  $80, $31, $72, $17, $f8	; log(2)
 
 
-floats_store_reg	.byte  0		; temp storage
 floats_temp_var         .byte  0,0,0,0,0        ; temporary storage for a float
 
 ub2float	.proc
 		; -- convert ubyte in SCRATCH_ZPB1 to float at address A/Y
-		;    clobbers A, Y
-		stx  P8ZP_SCRATCH_REG
+		;    clobbers A, X, Y
 		sta  P8ZP_SCRATCH_W2
 		sty  P8ZP_SCRATCH_W2+1
 		ldy  P8ZP_SCRATCH_B1
@@ -19,15 +17,12 @@ ub2float	.proc
 		jsr  GIVAYF
 _fac_to_mem	ldx  P8ZP_SCRATCH_W2
 		ldy  P8ZP_SCRATCH_W2+1
-		jsr  MOVMF
-		ldx  P8ZP_SCRATCH_REG
-		rts
+		jmp  MOVMF
 		.pend
 
 b2float		.proc
 		; -- convert byte in SCRATCH_ZPB1 to float at address A/Y
-		;    clobbers A, Y
-		stx  P8ZP_SCRATCH_REG
+		;    clobbers A, X, Y
 		sta  P8ZP_SCRATCH_W2
 		sty  P8ZP_SCRATCH_W2+1
 		lda  P8ZP_SCRATCH_B1
@@ -37,7 +32,7 @@ b2float		.proc
 
 uw2float	.proc
 		; -- convert uword in SCRATCH_ZPWORD1 to float at address A/Y
-		stx  P8ZP_SCRATCH_REG
+		;    clobbers X
 		sta  P8ZP_SCRATCH_W2
 		sty  P8ZP_SCRATCH_W2+1
 		lda  P8ZP_SCRATCH_W1
@@ -48,7 +43,7 @@ uw2float	.proc
 
 w2float		.proc
 		; -- convert word in SCRATCH_ZPWORD1 to float at address A/Y
-		stx  P8ZP_SCRATCH_REG
+		;    clobbers X
 		sta  P8ZP_SCRATCH_W2
 		sty  P8ZP_SCRATCH_W2+1
 		ldy  P8ZP_SCRATCH_W1
@@ -60,7 +55,7 @@ w2float		.proc
 
 cast_from_uw	.proc
 		; -- uword in A/Y into float var at (P8ZP_SCRATCH_W2)
-		stx  P8ZP_SCRATCH_REG
+		;    clobbers X
 		jsr  GIVUAYFAY
 		jmp  ub2float._fac_to_mem
 		.pend
@@ -68,7 +63,7 @@ cast_from_uw	.proc
 
 cast_from_w	.proc
 		; -- word in A/Y into float var at (P8ZP_SCRATCH_W2)
-		stx  P8ZP_SCRATCH_REG
+		;    clobbers X
 		jsr  GIVAYFAY
 		jmp  ub2float._fac_to_mem
 		.pend
@@ -76,7 +71,7 @@ cast_from_w	.proc
 
 cast_from_ub	.proc
 		; -- ubyte in Y into float var at (P8ZP_SCRATCH_W2)
-		stx  P8ZP_SCRATCH_REG
+		;    clobbers X
 		jsr  FREADUY
 		jmp  ub2float._fac_to_mem
 		.pend
@@ -84,169 +79,41 @@ cast_from_ub	.proc
 
 cast_from_b	.proc
 		; -- byte in A into float var at (P8ZP_SCRATCH_W2)
-		stx  P8ZP_SCRATCH_REG
+		;    clobbers X
 		jsr  FREADSA
 		jmp  ub2float._fac_to_mem
 		.pend
 
 cast_as_uw_into_ya	.proc               ; also used for float 2 ub
 		; -- cast float at A/Y to uword into Y/A
+		;    clobbers X
 		jsr  MOVFM
 		jmp  cast_FAC1_as_uw_into_ya
 		.pend
 
 cast_as_w_into_ay	.proc               ; also used for float 2 b
 		; -- cast float at A/Y to word into A/Y
+		;    clobbers X
 		jsr  MOVFM
 		jmp  cast_FAC1_as_w_into_ay
 		.pend
 
 cast_FAC1_as_uw_into_ya	.proc               ; also used for float 2 ub
 		; -- cast fac1 to uword into Y/A
-		stx  P8ZP_SCRATCH_REG
-		jsr  GETADR     ; into Y/A
-		ldx  P8ZP_SCRATCH_REG
-		rts
+		;    clobbers X
+		jmp  GETADR     ; into Y/A
 		.pend
 
 cast_FAC1_as_w_into_ay	.proc               ; also used for float 2 b
 		; -- cast fac1 to word into A/Y
-		stx  P8ZP_SCRATCH_REG
+		;    clobbers X
 		jsr  AYINT
 		ldy  floats.AYINT_facmo
 		lda  floats.AYINT_facmo+1
-		ldx  P8ZP_SCRATCH_REG
 		rts
 		.pend
 
 
-stack_b2float	.proc
-		; -- b2float operating on the stack
-		inx
-		lda  P8ESTACK_LO,x
-		stx  P8ZP_SCRATCH_REG
-		jsr  FREADSA
-		jmp  push_fac1._internal
-		.pend
-
-stack_w2float	.proc
-		; -- w2float operating on the stack
-		inx
-		ldy  P8ESTACK_LO,x
-		lda  P8ESTACK_HI,x
-		stx  P8ZP_SCRATCH_REG
-		jsr  GIVAYF
-		jmp  push_fac1._internal
-		.pend
-
-stack_ub2float	.proc
-		; -- ub2float operating on the stack
-		inx
-		lda  P8ESTACK_LO,x
-		stx  P8ZP_SCRATCH_REG
-		tay
-		lda  #0
-		jsr  GIVAYF
-		jmp  push_fac1._internal
-		.pend
-
-stack_uw2float	.proc
-		; -- uw2float operating on the stack
-		inx
-		lda  P8ESTACK_LO,x
-		ldy  P8ESTACK_HI,x
-		stx  P8ZP_SCRATCH_REG
-		jsr  GIVUAYFAY
-		jmp  push_fac1._internal
-		.pend
-
-stack_float2w	.proc               ; also used for float2b
-		jsr  pop_float_fac1
-		stx  P8ZP_SCRATCH_REG
-		jsr  AYINT
-		ldx  P8ZP_SCRATCH_REG
-		lda  floats.AYINT_facmo
-		sta  P8ESTACK_HI,x
-		lda  floats.AYINT_facmo+1
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
-
-stack_float2uw	.proc               ; also used for float2ub
-		jsr  pop_float_fac1
-		stx  P8ZP_SCRATCH_REG
-		jsr  GETADR
-		ldx  P8ZP_SCRATCH_REG
-		sta  P8ESTACK_HI,x
-		tya
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
-
-push_float	.proc
-		; ---- push mflpt5 in A/Y onto stack
-		; (taking 3 stack positions = 6 bytes of which 1 is padding)
-		sta  P8ZP_SCRATCH_W1
-		sty  P8ZP_SCRATCH_W1+1
-		ldy  #0
-		lda  (P8ZP_SCRATCH_W1),y
-		sta  P8ESTACK_LO,x
-		iny
-		lda  (P8ZP_SCRATCH_W1),y
-		sta  P8ESTACK_HI,x
-		dex
-		iny
-		lda  (P8ZP_SCRATCH_W1),y
-		sta  P8ESTACK_LO,x
-		iny
-		lda  (P8ZP_SCRATCH_W1),y
-		sta  P8ESTACK_HI,x
-		dex
-		iny
-		lda  (P8ZP_SCRATCH_W1),y
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
-
-pop_float	.proc
-		; ---- pops mflpt5 from stack to memory A/Y
-		; (frees 3 stack positions = 6 bytes of which 1 is padding)
-		sta  P8ZP_SCRATCH_W1
-		sty  P8ZP_SCRATCH_W1+1
-		ldy  #4
-		inx
-		lda  P8ESTACK_LO,x
-		sta  (P8ZP_SCRATCH_W1),y
-		dey
-		inx
-		lda  P8ESTACK_HI,x
-		sta  (P8ZP_SCRATCH_W1),y
-		dey
-		lda  P8ESTACK_LO,x
-		sta  (P8ZP_SCRATCH_W1),y
-		dey
-		inx
-		lda  P8ESTACK_HI,x
-		sta  (P8ZP_SCRATCH_W1),y
-		dey
-		lda  P8ESTACK_LO,x
-		sta  (P8ZP_SCRATCH_W1),y
-		rts
-		.pend
-
-pop_float_fac1	.proc
-		; -- pops float from stack into FAC1
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  pop_float
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jmp  MOVFM
-		.pend
-
 copy_float	.proc
 		; -- copies the 5 bytes of the mflt value pointed to by P8ZP_SCRATCH_W1,
 		;    into the 5 bytes pointed to by A/Y.  Clobbers A,Y.
@@ -272,25 +139,23 @@ copy_float	.proc
 
 inc_var_f	.proc
 		; -- add 1 to float pointed to by A/Y
+		;    clobbers X
 		sta  P8ZP_SCRATCH_W1
 		sty  P8ZP_SCRATCH_W1+1
-		stx  P8ZP_SCRATCH_REG
 		jsr  MOVFM
 		lda  #<FL_ONE_const
 		ldy  #>FL_ONE_const
 		jsr  FADD
 		ldx  P8ZP_SCRATCH_W1
 		ldy  P8ZP_SCRATCH_W1+1
-		jsr  MOVMF
-		ldx  P8ZP_SCRATCH_REG
-		rts
+		jmp  MOVMF
 		.pend
 
 dec_var_f	.proc
 		; -- subtract 1 from float pointed to by A/Y
+		;    clobbers X
 		sta  P8ZP_SCRATCH_W1
 		sty  P8ZP_SCRATCH_W1+1
-		stx  P8ZP_SCRATCH_REG
 		lda  #<FL_ONE_const
 		ldy  #>FL_ONE_const
 		jsr  MOVFM
@@ -299,23 +164,7 @@ dec_var_f	.proc
 		jsr  FSUB
 		ldx  P8ZP_SCRATCH_W1
 		ldy  P8ZP_SCRATCH_W1+1
-		jsr  MOVMF
-		ldx  P8ZP_SCRATCH_REG
-		rts
-		.pend
-
-
-pop_2_floats_f2_in_fac1	.proc
-		; -- pop 2 floats from stack, load the second one in FAC1 as well
-		lda  #<fmath_float2
-		ldy  #>fmath_float2
-		jsr  pop_float
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  pop_float
-		lda  #<fmath_float2
-		ldy  #>fmath_float2
-		jmp  MOVFM
+		jmp  MOVMF
 		.pend
 
 
@@ -323,71 +172,9 @@ fmath_float1	.byte 0,0,0,0,0	; storage for a mflpt5 value
 fmath_float2	.byte 0,0,0,0,0	; storage for a mflpt5 value
 
 
-push_fac1	.proc
-		; -- push the float in FAC1 onto the stack
-		stx  P8ZP_SCRATCH_REG
-_internal	ldx  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  MOVMF
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		ldx  P8ZP_SCRATCH_REG
-		jmp  push_float
-		.pend
-
-div_f		.proc
-		; -- push f1/f2 on stack
-		jsr  pop_2_floats_f2_in_fac1
-		stx  P8ZP_SCRATCH_REG
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  FDIV
-		jmp  push_fac1._internal
-		.pend
-
-add_f		.proc
-		; -- push f1+f2 on stack
-		jsr  pop_2_floats_f2_in_fac1
-		stx  P8ZP_SCRATCH_REG
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  FADD
-		jmp  push_fac1._internal
-		.pend
-
-sub_f		.proc
-		; -- push f1-f2 on stack
-		jsr  pop_2_floats_f2_in_fac1
-		stx  P8ZP_SCRATCH_REG
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  FSUB
-		jmp  push_fac1._internal
-		.pend
-
-mul_f		.proc
-		; -- push f1*f2 on stack
-		jsr  pop_2_floats_f2_in_fac1
-		stx  P8ZP_SCRATCH_REG
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  FMULT
-		jmp  push_fac1._internal
-		.pend
-
-neg_f		.proc
-		; -- toggle the sign bit on the stack
-		lda  P8ESTACK_HI+3,x
-		eor  #$80
-		sta  P8ESTACK_HI+3,x
-		rts
-		.pend
-
 var_fac1_less_f	.proc
-		; -- is the float in FAC1 < the variable AY?
-		stx  P8ZP_SCRATCH_REG
+		; -- is the float in FAC1 < the variable AY? Result in A. Clobbers X.
 		jsr  FCOMP
-		ldx  P8ZP_SCRATCH_REG
 		cmp  #255
 		beq  +
 		lda  #0
@@ -397,10 +184,8 @@ var_fac1_less_f	.proc
 		.pend
 
 var_fac1_lesseq_f	.proc
-		; -- is the float in FAC1 <= the variable AY?
-		stx  P8ZP_SCRATCH_REG
+		; -- is the float in FAC1 <= the variable AY?  Result in A. Clobbers X.
 		jsr  FCOMP
-		ldx  P8ZP_SCRATCH_REG
 		cmp  #0
 		beq  +
 		cmp  #255
@@ -412,10 +197,8 @@ var_fac1_lesseq_f	.proc
 		.pend
 
 var_fac1_greater_f	.proc
-		; -- is the float in FAC1 > the variable AY?
-		stx  P8ZP_SCRATCH_REG
+		; -- is the float in FAC1 > the variable AY?  Result in A. Clobbers X.
 		jsr  FCOMP
-		ldx  P8ZP_SCRATCH_REG
 		cmp  #1
 		beq  +
 		lda  #0
@@ -425,10 +208,8 @@ var_fac1_greater_f	.proc
 		.pend
 
 var_fac1_greatereq_f	.proc
-		; -- is the float in FAC1 >= the variable AY?
-		stx  P8ZP_SCRATCH_REG
+		; -- is the float in FAC1 >= the variable AY?  Result in A. Clobbers X.
 		jsr  FCOMP
-		ldx  P8ZP_SCRATCH_REG
 		cmp  #0
 		beq  +
 		cmp  #1
@@ -439,17 +220,23 @@ var_fac1_greatereq_f	.proc
 		rts
 		.pend
 
+var_fac1_equal_f	.proc
+		; -- are the floats numbers in FAC1 and the variable AY *not* identical?   Result in A. Clobbers X.
+		jsr  FCOMP
+		and  #1
+		eor  #1
+		rts
+		.pend
+
 var_fac1_notequal_f	.proc
-		; -- are the floats numbers in FAC1 and the variable AY *not* identical?
-		stx  P8ZP_SCRATCH_REG
+		; -- are the floats numbers in FAC1 and the variable AY *not* identical?   Result in A. Clobbers X.
 		jsr  FCOMP
-		ldx  P8ZP_SCRATCH_REG
 		and  #1
 		rts
 		.pend
 
 vars_equal_f	.proc
-		; -- are the mflpt5 numbers in P8ZP_SCRATCH_W1 and AY identical?
+		; -- are the mflpt5 numbers in P8ZP_SCRATCH_W1 and AY identical?  Result in A
 		sta  P8ZP_SCRATCH_W2
 		sty  P8ZP_SCRATCH_W2+1
 		ldy  #0
@@ -478,51 +265,13 @@ _false		lda  #0
 		rts
 		.pend
 
-equal_f		.proc
-		; -- are the two mflpt5 numbers on the stack identical?
-		inx
-		inx
-		inx
-		inx
-		lda  P8ESTACK_LO-3,x
-		cmp  P8ESTACK_LO,x
-		bne  _equals_false
-		lda  P8ESTACK_LO-2,x
-		cmp  P8ESTACK_LO+1,x
-		bne  _equals_false
-		lda  P8ESTACK_LO-1,x
-		cmp  P8ESTACK_LO+2,x
-		bne  _equals_false
-		lda  P8ESTACK_HI-2,x
-		cmp  P8ESTACK_HI+1,x
-		bne  _equals_false
-		lda  P8ESTACK_HI-1,x
-		cmp  P8ESTACK_HI+2,x
-		bne  _equals_false
-_equals_true	lda  #1
-_equals_store	inx
-		sta  P8ESTACK_LO+1,x
-		rts
-_equals_false	lda  #0
-		beq  _equals_store
-		.pend
-
-notequal_f	.proc
-		; -- are the two mflpt5 numbers on the stack different?
-		jsr  equal_f
-		eor  #1		; invert the result
-		sta  P8ESTACK_LO+1,x
-		rts
-		.pend
 
 vars_less_f	.proc
-		; -- is float in AY < float in P8ZP_SCRATCH_W2 ?
+		; -- is float in AY < float in P8ZP_SCRATCH_W2 ?   Result in A. Clobbers X.
 		jsr  MOVFM
 		lda  P8ZP_SCRATCH_W2
 		ldy  P8ZP_SCRATCH_W2+1
-		stx  P8ZP_SCRATCH_REG
 		jsr  FCOMP
-		ldx  P8ZP_SCRATCH_REG
 		cmp  #255
 		bne  +
 		lda  #1
@@ -532,13 +281,11 @@ vars_less_f	.proc
 		.pend
 
 vars_lesseq_f	.proc
-		; -- is float in AY <= float in P8ZP_SCRATCH_W2 ?
+		; -- is float in AY <= float in P8ZP_SCRATCH_W2 ?  Result in A. Clobbers X.
 		jsr  MOVFM
 		lda  P8ZP_SCRATCH_W2
 		ldy  P8ZP_SCRATCH_W2+1
-		stx  P8ZP_SCRATCH_REG
 		jsr  FCOMP
-		ldx  P8ZP_SCRATCH_REG
 		cmp  #255
 		bne  +
 -		lda  #1
@@ -550,7 +297,7 @@ vars_lesseq_f	.proc
 		.pend
 
 less_f		.proc
-		; -- is f1 < f2?
+		; -- is f1 < f2?    Result in A. Clobbers X.
 		jsr  compare_floats
 		cmp  #255
 		beq  compare_floats._return_true
@@ -559,7 +306,7 @@ less_f		.proc
 
 
 lesseq_f	.proc
-		; -- is f1 <= f2?
+		; -- is f1 <= f2?  Result in A. Clobbers X.
 		jsr  compare_floats
 		cmp  #255
 		beq  compare_floats._return_true
@@ -569,7 +316,7 @@ lesseq_f	.proc
 		.pend
 
 greater_f	.proc
-		; -- is f1 > f2?
+		; -- is f1 > f2?  Result in A. Clobbers X.
 		jsr  compare_floats
 		cmp  #1
 		beq  compare_floats._return_true
@@ -577,7 +324,7 @@ greater_f	.proc
 		.pend
 
 greatereq_f	.proc
-		; -- is f1 >= f2?
+		; -- is f1 >= f2?  Result in A. Clobbers X.
 		jsr  compare_floats
 		cmp  #1
 		beq  compare_floats._return_true
@@ -586,32 +333,9 @@ greatereq_f	.proc
 		bne  compare_floats._return_false
 		.pend
 
-compare_floats	.proc
-		lda  #<fmath_float2
-		ldy  #>fmath_float2
-		jsr  pop_float
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  pop_float
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  MOVFM		; fac1 = flt1
-		lda  #<fmath_float2
-		ldy  #>fmath_float2
-		stx  P8ZP_SCRATCH_REG
-		jsr  FCOMP		; A = flt1 compared with flt2 (0=equal, 1=flt1>flt2, 255=flt1<flt2)
-		ldx  P8ZP_SCRATCH_REG
-		rts
-_return_false	lda  #0
-_return_result  sta  P8ESTACK_LO,x
-		dex
-		rts
-_return_true	lda  #1
-		bne  _return_result
-		.pend
-
 set_array_float_from_fac1	.proc
 		; -- set the float in FAC1 in the array (index in A, array in P8ZP_SCRATCH_W1)
+		;    clobbers X
 		sta  P8ZP_SCRATCH_B1
 		asl  a
 		asl  a
@@ -622,11 +346,8 @@ set_array_float_from_fac1	.proc
 		adc  P8ZP_SCRATCH_W1
 		bcc  +
 		iny
-+		stx  floats_store_reg
-		tax
-		jsr  MOVMF
-		ldx  floats_store_reg
-		rts
++		tax
+		jmp  MOVMF
 		.pend
 
 
@@ -669,54 +390,63 @@ set_array_float		.proc
 		.pend
 
 
-equal_zero	.proc
-		jsr  floats.pop_float_fac1
-		jsr  floats.SIGN
-		beq  _true
-		bne  _false
-_true		lda  #1
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-_false		lda  #0
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
+pushFAC1    .proc
+	;-- push floating point in FAC onto the cpu stack
+	; save return address
+	pla
+	sta  P8ZP_SCRATCH_W2
+	pla
+	sta  P8ZP_SCRATCH_W2+1
+	ldx  #<floats.floats_temp_var
+	ldy  #>floats.floats_temp_var
+	jsr  floats.MOVMF
+	lda  floats.floats_temp_var
+	pha
+	lda  floats.floats_temp_var+1
+	pha
+	lda  floats.floats_temp_var+2
+	pha
+	lda  floats.floats_temp_var+3
+	pha
+	lda  floats.floats_temp_var+4
+	pha
+	; re-push return address
+	lda  P8ZP_SCRATCH_W2+1
+	pha
+	lda  P8ZP_SCRATCH_W2
+	pha
+	rts
+	.pend
 
-notequal_zero	.proc
-		jsr  floats.pop_float_fac1
-		jsr  floats.SIGN
-		bne  equal_zero._true
-		beq  equal_zero._false
-		.pend
+popFAC .proc
+	; -- pop floating point value from cpu stack into FAC1 or FAC2 (
+	;    carry flag clear=FAC1, carry set=FAC2
+	; save return address
+	pla
+	sta  P8ZP_SCRATCH_W2
+	pla
+	sta  P8ZP_SCRATCH_W2+1
+	pla
+	sta  floats.floats_temp_var+4
+	pla
+	sta  floats.floats_temp_var+3
+	pla
+	sta  floats.floats_temp_var+2
+	pla
+	sta  floats.floats_temp_var+1
+	pla
+	sta  floats.floats_temp_var
+	lda  #<floats.floats_temp_var
+	ldy  #>floats.floats_temp_var
+	bcs  +
+	jsr  floats.MOVFM
+	jmp  ++
++       jsr  floats.CONUPK
++	; re-push return address
+	lda  P8ZP_SCRATCH_W2+1
+	pha
+	lda  P8ZP_SCRATCH_W2
+	pha
+	rts
+	.pend
 
-greater_zero	.proc
-		jsr  floats.pop_float_fac1
-		jsr  floats.SIGN
-		beq  equal_zero._false
-		bpl  equal_zero._true
-		jmp  equal_zero._false
-		.pend
-
-less_zero	.proc
-		jsr  floats.pop_float_fac1
-		jsr  floats.SIGN
-		bmi  equal_zero._true
-		jmp  equal_zero._false
-		.pend
-
-greaterequal_zero	.proc
-		jsr  floats.pop_float_fac1
-		jsr  floats.SIGN
-		bpl  equal_zero._true
-		jmp  equal_zero._false
-		.pend
-
-lessequal_zero	.proc
-		jsr  floats.pop_float_fac1
-		jsr  floats.SIGN
-		beq  equal_zero._true
-		bmi  equal_zero._true
-		jmp  equal_zero._false
-		.pend

compiler/res/prog8lib/c64/floats.p8

@@ -5,6 +5,7 @@
 
 floats {
 	; ---- this block contains C-64 floating point related functions ----
+    %option no_symbol_prefixing
 
         const float  PI     = 3.141592653589793
         const float  TWOPI  = 6.283185307179586
@@ -45,7 +46,7 @@ romsub $b391 = GIVAYF(ubyte lo @ Y, ubyte hi @ A) clobbers(A,X,Y)
 
 romsub $b3a2 = FREADUY(ubyte value @ Y) clobbers(A,X,Y)     ; 8 bit unsigned Y -> float in fac1
 romsub $bc3c = FREADSA(byte value @ A) clobbers(A,X,Y)      ; 8 bit signed A -> float in fac1
-romsub $b7b5 = FREADSTR(ubyte length @ A) clobbers(A,X,Y)   ; str -> fac1, $22/23 must point to string, A=string length
+romsub $b7b5 = FREADSTR(ubyte length @ A) clobbers(A,X,Y)   ; str -> fac1, $22/23 must point to string, A=string length.  Also see parse_f()
 romsub $aabc = FPRINTLN() clobbers(A,X,Y)                   ; print string of fac1, on one line (= with newline) destroys fac1.  (consider FOUT + STROUT as well)
 romsub $bddd = FOUT() clobbers(X) -> uword @ AY             ; fac1 -> string, address returned in AY ($0100)
 
@@ -158,12 +159,19 @@ asmsub  GETADRAY  () clobbers(X) -> uword @ AY  {
 
 sub rndf() -> float {
     %asm {{
-        stx  P8ZP_SCRATCH_REG
         lda  #1
         jsr  FREADSA
-        jsr  RND		; rng into fac1
-        ldx  P8ZP_SCRATCH_REG
-        rts
+        jmp  RND		; rng into fac1
+    }}
+}
+
+asmsub parse_f(str value @AY) -> float @FAC1 {
+    %asm {{
+        sta  $22
+        sty  $23
+        jsr  prog8_lib.strlen
+        tya
+        jmp  FREADSTR
     }}
 }
 

compiler/res/prog8lib/c64/floats_funcs.asm

@@ -1,13 +1,6 @@
 ; --- floating point builtin functions
 
 
-func_sign_f_stack	.proc
-		jsr  func_sign_f_into_A
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
-
 func_sign_f_into_A	.proc
 		jsr  MOVFM
 		jmp  SIGN
@@ -146,17 +139,11 @@ func_all_f_stack	.proc
 		.pend
 
 func_abs_f_into_FAC1    .proc
-        stx  P8ZP_SCRATCH_REG
         jsr  MOVFM
-        jsr  ABS
-        ldx  P8ZP_SCRATCH_REG
-        rts
+        jmp  ABS
 	.pend
 
 func_sqrt_into_FAC1     .proc
-        stx  P8ZP_SCRATCH_REG
         jsr  MOVFM
-        jsr  SQR
-        ldx  P8ZP_SCRATCH_REG
-        rts
+        jmp  SQR
 	.pend

compiler/res/prog8lib/c64/graphics.p8

@@ -1,33 +1,42 @@
-%import textio
-
 ; bitmap pixel graphics module for the C64
 ; only black/white monochrome 320x200 for now
-; assumes bitmap screen memory is $2000-$3fff
+;
+; Sets character matrix and bitmap screen memory at a higher memory location $5c00-$7fff
+; so that the program itself can be larger without starting to overwrite the graphics memory.
 
 graphics {
-    const uword BITMAP_ADDRESS = $2000
+    %option no_symbol_prefixing
+
     const uword WIDTH = 320
     const ubyte HEIGHT = 200
 
+    const uword BITMAP_ADDRESS = $6000      ; MUST BE IN REGULAR RAM if you are not messing with ROM/RAM banking.
+    const uword CHARS_ADDRESS = $5c00       ; must be in same vic bank as the bitmap
+
     sub enable_bitmap_mode() {
         ; enable bitmap screen, erase it and set colors to black/white.
-        c64.SCROLY = %00111011
-        c64.SCROLX = %00001000
-        c64.VMCSB = (c64.VMCSB & %11110000) | %00001000   ; $2000-$3fff
         clear_screen(1, 0)
+        c64.SCROLY = %00111011      ; enable bitmap graphics mode
+        c64.SCROLX = %00001000      ; 40 column mode, no scrolling, multicolor bitmap off
+        c64.VMCSB = (lsb(CHARS_ADDRESS >> 6) & $F0) | (((BITMAP_ADDRESS & $3fff) / $0800) << 1)     ; set bitmap address
+        c64.CIA2DDRA |= %11
+        c64.CIA2PRA = lsb(BITMAP_ADDRESS >> 14) ^ 3     ; set VIC bank.
     }
 
     sub disable_bitmap_mode() {
-        ; enables text mode, erase the text screen, color white
-        c64.SCROLY = %00011011
-        c64.SCROLX = %00001000
-        c64.VMCSB = (c64.VMCSB & %11110000) | %00000100   ; $1000-$2fff
-        txt.fill_screen(' ', 1)
+        ; enables erase the text screen, text mode
+        sys.memset(CHARS_ADDRESS, 40*25, sc:' ')
+        c64.SCROLY = %00011011      ; disable bitmap graphics mode
+        c64.SCROLX = %00001000      ; 40 column mode, no scrolling
+        c64.VMCSB  = %00010100      ; screen addresses back to defaults
+        c64.CIA2DDRA |= %11
+        c64.CIA2PRA = %11   ; back to VIC bank 0.
     }
 
     sub clear_screen(ubyte pixelcolor, ubyte bgcolor) {
         sys.memset(BITMAP_ADDRESS, 320*200/8, 0)
-        txt.fill_screen(pixelcolor << 4 | bgcolor, 0)
+        sys.memset(CHARS_ADDRESS, 40*25, pixelcolor << 4 | bgcolor)
+        sys.memset(cbm.Colors, 40*25, 255)
     }
 
     sub line(uword @zp x1, ubyte @zp y1, uword @zp x2, ubyte @zp y2) {
@@ -125,46 +134,46 @@ graphics {
         }
     }
 
-    sub rect(uword x, ubyte y, uword width, ubyte height) {
+    sub rect(uword xx, ubyte yy, uword width, ubyte height) {
         if width==0 or height==0
             return
-        horizontal_line(x, y, width)
+        horizontal_line(xx, yy, width)
         if height==1
             return
-        horizontal_line(x, y+height-1, width)
-        vertical_line(x, y+1, height-2)
+        horizontal_line(xx, yy+height-1, width)
+        vertical_line(xx, yy+1, height-2)
         if width==1
             return
-        vertical_line(x+width-1, y+1, height-2)
+        vertical_line(xx+width-1, yy+1, height-2)
     }
 
-    sub fillrect(uword x, ubyte y, uword width, ubyte height) {
+    sub fillrect(uword xx, ubyte yy, uword width, ubyte height) {
         if width==0
             return
         repeat height {
-            horizontal_line(x, y, width)
-            y++
+            horizontal_line(xx, yy, width)
+            yy++
         }
     }
 
-    sub horizontal_line(uword x, ubyte y, uword length) {
+    sub horizontal_line(uword xx, ubyte yy, uword length) {
         if length<8 {
-            internal_plotx=x
+            internal_plotx=xx
             repeat lsb(length) {
-                internal_plot(y)
+                internal_plot(yy)
                 internal_plotx++
             }
             return
         }
 
-        ubyte separate_pixels = lsb(x) & 7
-        uword addr = get_y_lookup(y) + (x&$fff8)
+        ubyte separate_pixels = lsb(xx) & 7
+        uword pixaddr = get_y_lookup(yy) + (xx&$fff8)
 
         if separate_pixels {
             %asm {{
-                lda  addr
+                lda  pixaddr
                 sta  P8ZP_SCRATCH_W1
-                lda  addr+1
+                lda  pixaddr+1
                 sta  P8ZP_SCRATCH_W1+1
                 ldy  separate_pixels
                 lda  hline_filled_right,y
@@ -173,7 +182,7 @@ graphics {
                 ora  (P8ZP_SCRATCH_W1),y
                 sta  (P8ZP_SCRATCH_W1),y
             }}
-            addr += 8
+            pixaddr += 8
             length += separate_pixels
             length -= 8
         }
@@ -183,16 +192,15 @@ graphics {
                 lda  length
                 and  #7
                 sta  separate_pixels
-                stx  P8ZP_SCRATCH_REG
                 lsr  length+1
                 ror  length
                 lsr  length+1
                 ror  length
                 lsr  length+1
                 ror  length
-                lda  addr
+                lda  pixaddr
                 sta  _modified+1
-                lda  addr+1
+                lda  pixaddr+1
                 sta  _modified+2
                 lda  length
                 ora  length+1
@@ -208,8 +216,7 @@ _modified       stx  $ffff      ; modified
                 inc  _modified+2
 +               dey
                 bne  _modified
-_zero           ldx  P8ZP_SCRATCH_REG
-
+_zero
                 ldy  separate_pixels
                 beq  hline_zero2
                 lda  _modified+1
@@ -227,11 +234,11 @@ hline_zero2
         }
     }
 
-    sub vertical_line(uword x, ubyte y, ubyte height) {
-        internal_plotx = x
+    sub vertical_line(uword xx, ubyte yy, ubyte height) {
+        internal_plotx = xx
         repeat height {
-            internal_plot(y)
-            y++
+            internal_plot(yy)
+            yy++
         }
     }
 
@@ -300,8 +307,8 @@ hline_zero2
 ; here is the non-asm code for the plot routine below:
 ;    sub plot_nonasm(uword px, ubyte py) {
 ;        ubyte[] ormask = [128, 64, 32, 16, 8, 4, 2, 1]
-;        uword addr = BITMAP_ADDRESS + 320*(py>>3) + (py & 7) + (px & %0000000111111000)
-;        @(addr) |= ormask[lsb(px) & 7]
+;        uword pixaddr = BITMAP_ADDRESS + 320*(py>>3) + (py & 7) + (px & %0000000111111000)
+;        @(pixaddr) |= ormask[lsb(px) & 7]
 ;    }
 
     inline asmsub  plot(uword plotx @XY, ubyte ploty @A) clobbers (A, X, Y) {
@@ -352,7 +359,7 @@ _ormask     .byte 128, 64, 32, 16, 8, 4, 2, 1
 ; the y lookup tables encodes this formula:  BITMAP_ADDRESS + 320*(py>>3) + (py & 7)    (y from 0..199)
 ; We use the 64tass syntax for range expressions to calculate this table on assembly time.
 
-_plot_y_values := $2000 + 320*(range(200)>>3) + (range(200) & 7)
+_plot_y_values := BITMAP_ADDRESS + 320*(range(200)>>3) + (range(200) & 7)
 
 _y_lookup_lo    .byte  <_plot_y_values
 _y_lookup_hi    .byte  >_plot_y_values
@@ -360,7 +367,7 @@ _y_lookup_hi    .byte  >_plot_y_values
         }}
     }
 
-    asmsub get_y_lookup(ubyte y @Y) -> uword @AY {
+    asmsub get_y_lookup(ubyte yy @Y) -> uword @AY {
         %asm {{
             lda  internal_plot._y_lookup_lo,y
             pha

compiler/res/prog8lib/c64/syslib.p8

@@ -6,6 +6,8 @@ cbm {
 
     ; Commodore (CBM) common variables, vectors and kernal routines
 
+    %option no_symbol_prefixing
+
         &ubyte  TIME_HI         = $a0       ; software jiffy clock, hi byte
         &ubyte  TIME_MID        = $a1       ;  .. mid byte
         &ubyte  TIME_LO         = $a2       ;    .. lo byte. Updated by IRQ every 1/60 sec
@@ -89,7 +91,7 @@ romsub $FFC6 = CHKIN(ubyte logical @ X) clobbers(A,X) -> bool @Pc    ; (via 798
 romsub $FFC9 = CHKOUT(ubyte logical @ X) clobbers(A,X)          ; (via 800 ($320)) define an output channel
 romsub $FFCC = CLRCHN() clobbers(A,X)                           ; (via 802 ($322)) restore default devices
 romsub $FFCF = CHRIN() clobbers(X, Y) -> ubyte @ A   ; (via 804 ($324)) input a character (for keyboard, read a whole line from the screen) A=byte read.
-romsub $FFD2 = CHROUT(ubyte char @ A)                           ; (via 806 ($326)) output a character
+romsub $FFD2 = CHROUT(ubyte character @ A)                      ; (via 806 ($326)) output a character
 romsub $FFD5 = LOAD(ubyte verify @ A, uword address @ XY) -> bool @Pc, ubyte @ A, uword @ XY     ; (via 816 ($330)) load from device
 romsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) -> bool @ Pc, ubyte @ A          ; (via 818 ($332)) save to a device
 romsub $FFDB = SETTIM(ubyte low @ A, ubyte middle @ X, ubyte high @ Y)      ; set the software clock
@@ -102,34 +104,26 @@ romsub $FFED = SCREEN() -> ubyte @ X, ubyte @ Y                 ; read number of
 romsub $FFF0 = PLOT(ubyte col @ Y, ubyte row @ X, bool dir @ Pc) -> ubyte @ X, ubyte @ Y       ; read/set position of cursor on screen.  Use txt.plot for a 'safe' wrapper that preserves X.
 romsub $FFF3 = IOBASE() -> uword @ XY                           ; read base address of I/O devices
 
-asmsub STOP2() -> ubyte @A  {
+asmsub STOP2() clobbers(X) -> ubyte @A  {
     ; -- check if STOP key was pressed, returns true if so.  More convenient to use than STOP() because that only sets the carry status flag.
     %asm {{
-        txa
-        pha
         jsr  cbm.STOP
         beq  +
-        pla
-        tax
         lda  #0
         rts
-+       pla
-        tax
-        lda  #1
++       lda  #1
         rts
     }}
 }
 
-asmsub RDTIM16() -> uword @AY {
+asmsub RDTIM16() clobbers(X) -> uword @AY {
     ; --  like RDTIM() but only returning the lower 16 bits in AY for convenience
     %asm {{
-        stx  P8ZP_SCRATCH_REG
         jsr  cbm.RDTIM
         pha
         txa
         tay
         pla
-        ldx  P8ZP_SCRATCH_REG
         rts
     }}
 }
@@ -139,7 +133,10 @@ asmsub RDTIM16() -> uword @AY {
 c64 {
         ; C64 I/O registers (VIC, SID, CIA)
 
+    %option no_symbol_prefixing
+
         ; the default locations of the 8 sprite pointers (store address of sprite / 64)
+        ; (depending on the VIC bank and screen ram address selection these can be shifted around though)
         &ubyte  SPRPTR0         = 2040
         &ubyte  SPRPTR1         = 2041
         &ubyte  SPRPTR2         = 2042
@@ -293,6 +290,8 @@ c64 {
 sys {
     ; ------- lowlevel system routines --------
 
+    %option no_symbol_prefixing
+
     const ubyte target = 64         ;  compilation target specifier.  64 = C64, 128 = C128, 16 = CommanderX16.
 
 
@@ -302,7 +301,7 @@ asmsub  init_system()  {
     ; This means that the BASIC, KERNAL and CHARGEN ROMs are banked in,
     ; the VIC, SID and CIA chips are reset, screen is cleared, and the default IRQ is set.
     ; Also a different color scheme is chosen to identify ourselves a little.
-    ; Uppercase charset is activated, and all three registers set to 0, status flags cleared.
+    ; Uppercase charset is activated.
     %asm {{
         sei
         cld
@@ -407,10 +406,6 @@ _irq_handler_init
 		sta  IRQ_SCRATCH_ZPWORD2
 		lda  P8ZP_SCRATCH_W2+1
 		sta  IRQ_SCRATCH_ZPWORD2+1
-		; Set X to the bottom 32 bytes of the evaluation stack, to HOPEFULLY not clobber it.
-		; This leaves 128-32=96 stack entries for the main program, and 32 stack entries for the IRQ handler.
-		; We assume IRQ handlers don't contain complex expressions taking up more than that.
-		ldx  #32
 		cld
 		rts
 
@@ -732,111 +727,139 @@ _longcopy
 
 cx16 {
 
+    %option no_symbol_prefixing
+
     ; the sixteen virtual 16-bit registers that the CX16 has defined in the zeropage
     ; they are simulated on the C64 as well but their location in memory is different
-    ; (because there's no room for them in the zeropage)
-    ; they are allocated at the bottom of the eval-stack (should be ample space unless
-    ; you're doing insane nesting of expressions...)
-    ; NOTE: the memory location of these registers can change based on the "-esa" compiler option
-    &uword r0  = $cf00
-    &uword r1  = $cf02
-    &uword r2  = $cf04
-    &uword r3  = $cf06
-    &uword r4  = $cf08
-    &uword r5  = $cf0a
-    &uword r6  = $cf0c
-    &uword r7  = $cf0e
-    &uword r8  = $cf10
-    &uword r9  = $cf12
-    &uword r10 = $cf14
-    &uword r11 = $cf16
-    &uword r12 = $cf18
-    &uword r13 = $cf1a
-    &uword r14 = $cf1c
-    &uword r15 = $cf1e
+    ; (because there's no room for them in the zeropage in the default configuration)
+    ; Note that when using ZP options that free up more of the zeropage (such as %zeropage kernalsafe)
+    ; there might be enough space to put them there after all, and the compiler will change these addresses!
+    &uword r0  = $cfe0
+    &uword r1  = $cfe2
+    &uword r2  = $cfe4
+    &uword r3  = $cfe6
+    &uword r4  = $cfe8
+    &uword r5  = $cfea
+    &uword r6  = $cfec
+    &uword r7  = $cfee
+    &uword r8  = $cff0
+    &uword r9  = $cff2
+    &uword r10 = $cff4
+    &uword r11 = $cff6
+    &uword r12 = $cff8
+    &uword r13 = $cffa
+    &uword r14 = $cffc
+    &uword r15 = $cffe
 
-    &word r0s  = $cf00
-    &word r1s  = $cf02
-    &word r2s  = $cf04
-    &word r3s  = $cf06
-    &word r4s  = $cf08
-    &word r5s  = $cf0a
-    &word r6s  = $cf0c
-    &word r7s  = $cf0e
-    &word r8s  = $cf10
-    &word r9s  = $cf12
-    &word r10s = $cf14
-    &word r11s = $cf16
-    &word r12s = $cf18
-    &word r13s = $cf1a
-    &word r14s = $cf1c
-    &word r15s = $cf1e
+    &word r0s  = $cfe0
+    &word r1s  = $cfe2
+    &word r2s  = $cfe4
+    &word r3s  = $cfe6
+    &word r4s  = $cfe8
+    &word r5s  = $cfea
+    &word r6s  = $cfec
+    &word r7s  = $cfee
+    &word r8s  = $cff0
+    &word r9s  = $cff2
+    &word r10s = $cff4
+    &word r11s = $cff6
+    &word r12s = $cff8
+    &word r13s = $cffa
+    &word r14s = $cffc
+    &word r15s = $cffe
 
-    &ubyte r0L  = $cf00
-    &ubyte r1L  = $cf02
-    &ubyte r2L  = $cf04
-    &ubyte r3L  = $cf06
-    &ubyte r4L  = $cf08
-    &ubyte r5L  = $cf0a
-    &ubyte r6L  = $cf0c
-    &ubyte r7L  = $cf0e
-    &ubyte r8L  = $cf10
-    &ubyte r9L  = $cf12
-    &ubyte r10L = $cf14
-    &ubyte r11L = $cf16
-    &ubyte r12L = $cf18
-    &ubyte r13L = $cf1a
-    &ubyte r14L = $cf1c
-    &ubyte r15L = $cf1e
+    &ubyte r0L  = $cfe0
+    &ubyte r1L  = $cfe2
+    &ubyte r2L  = $cfe4
+    &ubyte r3L  = $cfe6
+    &ubyte r4L  = $cfe8
+    &ubyte r5L  = $cfea
+    &ubyte r6L  = $cfec
+    &ubyte r7L  = $cfee
+    &ubyte r8L  = $cff0
+    &ubyte r9L  = $cff2
+    &ubyte r10L = $cff4
+    &ubyte r11L = $cff6
+    &ubyte r12L = $cff8
+    &ubyte r13L = $cffa
+    &ubyte r14L = $cffc
+    &ubyte r15L = $cffe
 
-    &ubyte r0H  = $cf01
-    &ubyte r1H  = $cf03
-    &ubyte r2H  = $cf05
-    &ubyte r3H  = $cf07
-    &ubyte r4H  = $cf09
-    &ubyte r5H  = $cf0b
-    &ubyte r6H  = $cf0d
-    &ubyte r7H  = $cf0f
-    &ubyte r8H  = $cf11
-    &ubyte r9H  = $cf13
-    &ubyte r10H = $cf15
-    &ubyte r11H = $cf17
-    &ubyte r12H = $cf19
-    &ubyte r13H = $cf1b
-    &ubyte r14H = $cf1d
-    &ubyte r15H = $cf1f
+    &ubyte r0H  = $cfe1
+    &ubyte r1H  = $cfe3
+    &ubyte r2H  = $cfe5
+    &ubyte r3H  = $cfe7
+    &ubyte r4H  = $cfe9
+    &ubyte r5H  = $cfeb
+    &ubyte r6H  = $cfed
+    &ubyte r7H  = $cfef
+    &ubyte r8H  = $cff1
+    &ubyte r9H  = $cff3
+    &ubyte r10H = $cff5
+    &ubyte r11H = $cff7
+    &ubyte r12H = $cff9
+    &ubyte r13H = $cffb
+    &ubyte r14H = $cffd
+    &ubyte r15H = $cfff
 
-    &byte r0sL  = $cf00
-    &byte r1sL  = $cf02
-    &byte r2sL  = $cf04
-    &byte r3sL  = $cf06
-    &byte r4sL  = $cf08
-    &byte r5sL  = $cf0a
-    &byte r6sL  = $cf0c
-    &byte r7sL  = $cf0e
-    &byte r8sL  = $cf10
-    &byte r9sL  = $cf12
-    &byte r10sL = $cf14
-    &byte r11sL = $cf16
-    &byte r12sL = $cf18
-    &byte r13sL = $cf1a
-    &byte r14sL = $cf1c
-    &byte r15sL = $cf1e
+    &byte r0sL  = $cfe0
+    &byte r1sL  = $cfe2
+    &byte r2sL  = $cfe4
+    &byte r3sL  = $cfe6
+    &byte r4sL  = $cfe8
+    &byte r5sL  = $cfea
+    &byte r6sL  = $cfec
+    &byte r7sL  = $cfee
+    &byte r8sL  = $cff0
+    &byte r9sL  = $cff2
+    &byte r10sL = $cff4
+    &byte r11sL = $cff6
+    &byte r12sL = $cff8
+    &byte r13sL = $cffa
+    &byte r14sL = $cffc
+    &byte r15sL = $cffe
+
+    &byte r0sH  = $cfe1
+    &byte r1sH  = $cfe3
+    &byte r2sH  = $cfe5
+    &byte r3sH  = $cfe7
+    &byte r4sH  = $cfe9
+    &byte r5sH  = $cfeb
+    &byte r6sH  = $cfed
+    &byte r7sH  = $cfef
+    &byte r8sH  = $cff1
+    &byte r9sH  = $cff3
+    &byte r10sH = $cff5
+    &byte r11sH = $cff7
+    &byte r12sH = $cff9
+    &byte r13sH = $cffb
+    &byte r14sH = $cffd
+    &byte r15sH = $cfff
+
+    asmsub save_virtual_registers() clobbers(A,Y) {
+        %asm {{
+            ldy  #31
+    -       lda  cx16.r0,y
+            sta  _cx16_vreg_storage,y
+            dey
+            bpl  -
+            rts
+
+    _cx16_vreg_storage
+            .word 0,0,0,0,0,0,0,0
+            .word 0,0,0,0,0,0,0,0
+        }}
+    }
+
+    asmsub restore_virtual_registers() clobbers(A,Y) {
+        %asm {{
+            ldy  #31
+    -       lda  save_virtual_registers._cx16_vreg_storage,y
+            sta  cx16.r0,y
+            dey
+            bpl  -
+            rts
+        }}
+    }
 
-    &byte r0sH  = $cf01
-    &byte r1sH  = $cf03
-    &byte r2sH  = $cf05
-    &byte r3sH  = $cf07
-    &byte r4sH  = $cf09
-    &byte r5sH  = $cf0b
-    &byte r6sH  = $cf0d
-    &byte r7sH  = $cf0f
-    &byte r8sH  = $cf11
-    &byte r9sH  = $cf13
-    &byte r10sH = $cf15
-    &byte r11sH = $cf17
-    &byte r12sH = $cf19
-    &byte r13sH = $cf1b
-    &byte r14sH = $cf1d
-    &byte r15sH = $cf1f
 }

compiler/res/prog8lib/c64/textio.p8

@@ -6,6 +6,8 @@
 
 txt {
 
+    %option no_symbol_prefixing
+
 const ubyte DEFAULT_WIDTH = 40
 const ubyte DEFAULT_HEIGHT = 25
 
@@ -37,7 +39,7 @@ asmsub column(ubyte col @A) clobbers(A, X, Y) {
     }}
 }
 
-asmsub  fill_screen (ubyte char @ A, ubyte color @ Y) clobbers(A)  {
+asmsub  fill_screen (ubyte character @ A, ubyte color @ Y) clobbers(A)  {
 	; ---- fill the character screen with the given fill character and character color.
 	;      (assumes screen and color matrix are at their default addresses)
 
@@ -52,7 +54,7 @@ asmsub  fill_screen (ubyte char @ A, ubyte color @ Y) clobbers(A)  {
 
 }
 
-asmsub  clear_screenchars (ubyte char @ A) clobbers(Y)  {
+asmsub  clear_screenchars (ubyte character @ A) clobbers(Y)  {
 	; ---- clear the character screen with the given fill character (leaves colors)
 	;      (assumes screen matrix is at the default address)
 	%asm {{
@@ -94,13 +96,12 @@ sub uppercase() {
     c64.VMCSB &= ~2
 }
 
-asmsub  scroll_left  (bool alsocolors @ Pc) clobbers(A, Y)  {
+asmsub  scroll_left  (bool alsocolors @ Pc) clobbers(A, X, Y)  {
 	; ---- scroll the whole screen 1 character to the left
 	;      contents of the rightmost column are unchanged, you should clear/refill this yourself
 	;      Carry flag determines if screen color data must be scrolled too
 
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		bcc _scroll_screen
 
 +               ; scroll the screen and the color memory
@@ -130,17 +131,15 @@ _scroll_screen  ; scroll only the screen memory
 		dey
 		bpl  -
 
-		ldx  P8ZP_SCRATCH_REG
 		rts
 	}}
 }
 
-asmsub  scroll_right  (bool alsocolors @ Pc) clobbers(A)  {
+asmsub  scroll_right  (bool alsocolors @ Pc) clobbers(A,X)  {
 	; ---- scroll the whole screen 1 character to the right
 	;      contents of the leftmost column are unchanged, you should clear/refill this yourself
 	;      Carry flag determines if screen color data must be scrolled too
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		bcc  _scroll_screen
 
 +               ; scroll the screen and the color memory
@@ -166,17 +165,15 @@ _scroll_screen  ; scroll only the screen memory
 		dex
 		bpl  -
 
-		ldx  P8ZP_SCRATCH_REG
 		rts
 	}}
 }
 
-asmsub  scroll_up  (bool alsocolors @ Pc) clobbers(A)  {
+asmsub  scroll_up  (bool alsocolors @ Pc) clobbers(A,X)  {
 	; ---- scroll the whole screen 1 character up
 	;      contents of the bottom row are unchanged, you should refill/clear this yourself
 	;      Carry flag determines if screen color data must be scrolled too
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		bcc  _scroll_screen
 
 +               ; scroll the screen and the color memory
@@ -202,17 +199,15 @@ _scroll_screen  ; scroll only the screen memory
 		dex
 		bpl  -
 
-		ldx  P8ZP_SCRATCH_REG
 		rts
 	}}
 }
 
-asmsub  scroll_down  (bool alsocolors @ Pc) clobbers(A)  {
+asmsub  scroll_down  (bool alsocolors @ Pc) clobbers(A,X)  {
 	; ---- scroll the whole screen 1 character down
 	;      contents of the top row are unchanged, you should refill/clear this yourself
 	;      Carry flag determines if screen color data must be scrolled too
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		bcc  _scroll_screen
 
 +               ; scroll the screen and the color memory
@@ -238,12 +233,11 @@ _scroll_screen  ; scroll only the screen memory
 		dex
 		bpl  -
 
-		ldx  P8ZP_SCRATCH_REG
 		rts
 	}}
 }
 
-romsub $FFD2 = chrout(ubyte char @ A)    ; for consistency. You can also use cbm.CHROUT directly ofcourse.
+romsub $FFD2 = chrout(ubyte character @ A)    ; for consistency. You can also use cbm.CHROUT directly ofcourse.
 
 asmsub  print (str text @ AY) clobbers(A,Y)  {
 	; ---- print null terminated string from A/Y
@@ -263,10 +257,9 @@ asmsub  print (str text @ AY) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  print_ub0  (ubyte value @ A) clobbers(A,Y)  {
+asmsub  print_ub0  (ubyte value @ A) clobbers(A,X,Y)  {
 	; ---- print the ubyte in A in decimal form, with left padding 0s (3 positions total)
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		jsr  conv.ubyte2decimal
 		pha
 		tya
@@ -274,16 +267,13 @@ asmsub  print_ub0  (ubyte value @ A) clobbers(A,Y)  {
 		pla
 		jsr  cbm.CHROUT
 		txa
-		jsr  cbm.CHROUT
-		ldx  P8ZP_SCRATCH_REG
-		rts
+		jmp  cbm.CHROUT
 	}}
 }
 
-asmsub  print_ub  (ubyte value @ A) clobbers(A,Y)  {
+asmsub  print_ub  (ubyte value @ A) clobbers(A,X,Y)  {
 	; ---- print the ubyte in A in decimal form, without left padding 0s
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		jsr  conv.ubyte2decimal
 _print_byte_digits
 		pha
@@ -299,16 +289,13 @@ _print_byte_digits
         beq  _ones
         jsr  cbm.CHROUT
 _ones   txa
-		jsr  cbm.CHROUT
-		ldx  P8ZP_SCRATCH_REG
-		rts
+		jmp  cbm.CHROUT
 	}}
 }
 
-asmsub  print_b  (byte value @ A) clobbers(A,Y)  {
+asmsub  print_b  (byte value @ A) clobbers(A,X,Y)  {
 	; ---- print the byte in A in decimal form, without left padding 0s
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		pha
 		cmp  #0
 		bpl  +
@@ -320,10 +307,9 @@ asmsub  print_b  (byte value @ A) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  print_ubhex  (ubyte value @ A, bool prefix @ Pc) clobbers(A,Y)  {
+asmsub  print_ubhex  (ubyte value @ A, bool prefix @ Pc) clobbers(A,X,Y)  {
 	; ---- print the ubyte in A in hex form (if Carry is set, a radix prefix '$' is printed as well)
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		bcc  +
 		pha
 		lda  #'$'
@@ -332,16 +318,13 @@ asmsub  print_ubhex  (ubyte value @ A, bool prefix @ Pc) clobbers(A,Y)  {
 +		jsr  conv.ubyte2hex
 		jsr  cbm.CHROUT
 		tya
-		jsr  cbm.CHROUT
-		ldx  P8ZP_SCRATCH_REG
-		rts
+		jmp  cbm.CHROUT
 	}}
 }
 
-asmsub  print_ubbin  (ubyte value @ A, bool prefix @ Pc) clobbers(A,Y)  {
+asmsub  print_ubbin  (ubyte value @ A, bool prefix @ Pc) clobbers(A,X,Y)  {
 	; ---- print the ubyte in A in binary form (if Carry is set, a radix prefix '%' is printed as well)
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		sta  P8ZP_SCRATCH_B1
 		bcc  +
 		lda  #'%'
@@ -354,12 +337,11 @@ asmsub  print_ubbin  (ubyte value @ A, bool prefix @ Pc) clobbers(A,Y)  {
 +		jsr  cbm.CHROUT
 		dey
 		bne  -
-		ldx  P8ZP_SCRATCH_REG
 		rts
 	}}
 }
 
-asmsub  print_uwbin  (uword value @ AY, bool prefix @ Pc) clobbers(A,Y)  {
+asmsub  print_uwbin  (uword value @ AY, bool prefix @ Pc) clobbers(A,X,Y)  {
 	; ---- print the uword in A/Y in binary form (if Carry is set, a radix prefix '%' is printed as well)
 	%asm {{
 		pha
@@ -371,7 +353,7 @@ asmsub  print_uwbin  (uword value @ AY, bool prefix @ Pc) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  print_uwhex  (uword value @ AY, bool prefix @ Pc) clobbers(A,Y)  {
+asmsub  print_uwhex  (uword value @ AY, bool prefix @ Pc) clobbers(A,X,Y)  {
 	; ---- print the uword in A/Y in hexadecimal form (4 digits)
 	;      (if Carry is set, a radix prefix '$' is printed as well)
 	%asm {{
@@ -384,10 +366,9 @@ asmsub  print_uwhex  (uword value @ AY, bool prefix @ Pc) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  print_uw0  (uword value @ AY) clobbers(A,Y)  {
+asmsub  print_uw0  (uword value @ AY) clobbers(A,X,Y)  {
 	; ---- print the uword in A/Y in decimal form, with left padding 0s (5 positions total)
 	%asm {{
-	    stx  P8ZP_SCRATCH_REG
 		jsr  conv.uword2decimal
 		ldy  #0
 -		lda  conv.uword2decimal.decTenThousands,y
@@ -395,17 +376,14 @@ asmsub  print_uw0  (uword value @ AY) clobbers(A,Y)  {
 		jsr  cbm.CHROUT
 		iny
 		bne  -
-+		ldx  P8ZP_SCRATCH_REG
-		rts
++		rts
 	}}
 }
 
-asmsub  print_uw  (uword value @ AY) clobbers(A,Y)  {
+asmsub  print_uw  (uword value @ AY) clobbers(A,X,Y)  {
 	; ---- print the uword in A/Y in decimal form, without left padding 0s
 	%asm {{
-	    stx  P8ZP_SCRATCH_REG
 		jsr  conv.uword2decimal
-		ldx  P8ZP_SCRATCH_REG
 		ldy  #0
 -		lda  conv.uword2decimal.decTenThousands,y
 		beq  _allzero
@@ -426,7 +404,7 @@ _allzero
 	}}
 }
 
-asmsub  print_w  (word value @ AY) clobbers(A,Y)  {
+asmsub  print_w  (word value @ AY) clobbers(A,X,Y)  {
 	; ---- print the (signed) word in A/Y in decimal form, without left padding 0's
 	%asm {{
 		cpy  #0
@@ -487,7 +465,7 @@ asmsub  setchr  (ubyte col @X, ubyte row @Y, ubyte character @A) clobbers(A, Y)
 _mod		sta  $ffff		; modified
 		rts
 
-_screenrows	.word  $0400 + range(0, 1000, 40)
+_screenrows	.word  cbm.Screen + range(0, 1000, 40)
 	}}
 }
 
@@ -554,7 +532,7 @@ _mod		lda  $ffff		; modified
 	}}
 }
 
-sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
+sub  setcc  (ubyte col, ubyte row, ubyte character, ubyte charcolor)  {
 	; ---- set char+color at the given position on the screen
 	%asm {{
 		lda  row
@@ -566,13 +544,13 @@ sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
 		sta  _colormod+2
 		lda  setchr._screenrows,y
 		clc
-		adc  column
+		adc  col
 		sta  _charmod+1
 		sta  _colormod+1
 		bcc  +
 		inc  _charmod+2
 		inc  _colormod+2
-+		lda  char
++		lda  character
 _charmod	sta  $ffff		; modified
 		lda  charcolor
 _colormod	sta  $ffff		; modified
@@ -580,15 +558,10 @@ _colormod	sta  $ffff		; modified
 	}}
 }
 
-asmsub  plot  (ubyte col @ Y, ubyte row @ A) clobbers(A) {
-	; ---- safe wrapper around PLOT kernal routine, to save the X register.
+asmsub  plot  (ubyte col @ Y, ubyte row @ X) {
 	%asm  {{
-		stx  P8ZP_SCRATCH_REG
-		tax
 		clc
-		jsr  cbm.PLOT
-		ldx  P8ZP_SCRATCH_REG
-		rts
+		jmp  cbm.PLOT
 	}}
 }
 

compiler/res/prog8lib/conv.p8

@@ -2,29 +2,28 @@
 
 conv {
 
+    %option no_symbol_prefixing
+
 ; ----- number conversions to decimal strings ----
 
     str  @shared string_out = "????????????????"       ; result buffer for the string conversion routines
 
-asmsub  str_ub0  (ubyte value @ A) clobbers(A,Y)  {
+asmsub  str_ub0  (ubyte value @ A) clobbers(A,X,Y)  {
 	; ---- convert the ubyte in A in decimal string form, with left padding 0s (3 positions total)
 	%asm {{
-            stx  P8ZP_SCRATCH_REG
             jsr  conv.ubyte2decimal
             sty  string_out
             sta  string_out+1
             stx  string_out+2
             lda  #0
             sta  string_out+3
-            ldx  P8ZP_SCRATCH_REG
             rts
 	}}
 }
 
-asmsub  str_ub  (ubyte value @ A) clobbers(A,Y)  {
+asmsub  str_ub  (ubyte value @ A) clobbers(A,X,Y)  {
 	; ---- convert the ubyte in A in decimal string form, without left padding 0s
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		ldy  #0
 		sty  P8ZP_SCRATCH_B1
 		jsr  conv.ubyte2decimal
@@ -50,15 +49,13 @@ _output_byte_digits
                 iny
                 lda  #0
                 sta  string_out,y
-                ldx  P8ZP_SCRATCH_REG
                 rts
 	}}
 }
 
-asmsub  str_b  (byte value @ A) clobbers(A,Y)  {
+asmsub  str_b  (byte value @ A) clobbers(A,X,Y)  {
 	; ---- convert the byte in A in decimal string form, without left padding 0s
 	%asm {{
-            stx  P8ZP_SCRATCH_REG
             ldy  #0
             sty  P8ZP_SCRATCH_B1
             cmp  #0
@@ -73,7 +70,7 @@ asmsub  str_b  (byte value @ A) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  str_ubhex  (ubyte value @ A) clobbers(A,Y)  {
+asmsub  str_ubhex  (ubyte value @ A) clobbers(A,X,Y)  {
 	; ---- convert the ubyte in A in hex string form
 	%asm {{
             jsr  conv.ubyte2hex
@@ -85,7 +82,7 @@ asmsub  str_ubhex  (ubyte value @ A) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  str_ubbin  (ubyte value @ A) clobbers(A,Y)  {
+asmsub  str_ubbin  (ubyte value @ A) clobbers(A,X,Y)  {
 	; ---- convert the ubyte in A in binary string form
 	%asm {{
 	    sta  P8ZP_SCRATCH_B1
@@ -104,7 +101,7 @@ _digit      sta  string_out,y
 	}}
 }
 
-asmsub  str_uwbin  (uword value @ AY) clobbers(A,Y)  {
+asmsub  str_uwbin  (uword value @ AY) clobbers(A,X,Y)  {
 	; ---- convert the uword in A/Y in binary string form
 	%asm {{
 	    sta  P8ZP_SCRATCH_REG
@@ -143,10 +140,9 @@ asmsub  str_uwhex  (uword value @ AY) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  str_uw0  (uword value @ AY) clobbers(A,Y)  {
+asmsub  str_uw0  (uword value @ AY) clobbers(A,X,Y)  {
 	; ---- convert the uword in A/Y in decimal string form, with left padding 0s (5 positions total)
 	%asm {{
-	    stx  P8ZP_SCRATCH_REG
 	    jsr  conv.uword2decimal
 	    ldy  #0
 -           lda  conv.uword2decimal.decTenThousands,y
@@ -154,15 +150,13 @@ asmsub  str_uw0  (uword value @ AY) clobbers(A,Y)  {
             beq  +
             iny
             bne  -
-+           ldx  P8ZP_SCRATCH_REG
-	    rts
++	    rts
 	}}
 }
 
-asmsub  str_uw  (uword value @ AY) clobbers(A,Y)  {
+asmsub  str_uw  (uword value @ AY) clobbers(A,X,Y)  {
 	; ---- convert the uword in A/Y in decimal string form, without left padding 0s
 	%asm {{
-	    stx  P8ZP_SCRATCH_REG
 	    jsr  conv.uword2decimal
 	    ldx  #0
 _output_digits
@@ -180,7 +174,6 @@ _gotdigit   sta  string_out,x
             bne  _gotdigit
 _end        lda  #0
             sta  string_out,x
-            ldx  P8ZP_SCRATCH_REG
             rts
 
 _allzero    lda  #'0'
@@ -190,12 +183,11 @@ _allzero    lda  #'0'
 	}}
 }
 
-asmsub  str_w  (word value @ AY) clobbers(A,Y)  {
+asmsub  str_w  (word value @ AY) clobbers(A,X,Y)  {
 	; ---- convert the (signed) word in A/Y in decimal string form, without left padding 0's
 	%asm {{
 	    cpy  #0
 	    bpl  str_uw
-	    stx  P8ZP_SCRATCH_REG
 	    pha
 	    lda  #'-'
 	    sta  string_out
@@ -506,8 +498,7 @@ asmsub  ubyte2decimal  (ubyte value @A) -> ubyte @Y, ubyte @A, ubyte @X  {
 	; ---- A to decimal string in Y/A/X  (100s in Y, 10s in A, 1s in X)
 	%asm {{
 		ldy  #uword2decimal.ASCII_0_OFFSET
-		bne  uword2decimal.hex_try200
-		rts
+		jmp  uword2decimal.hex_try200
 	}}
 }
 
@@ -699,10 +690,9 @@ asmsub  byte2decimal  (byte value @A) -> ubyte @Y, ubyte @A, ubyte @X  {
 	}}
 }
 
-asmsub  ubyte2hex  (ubyte value @A) -> ubyte @A, ubyte @Y  {
+asmsub  ubyte2hex  (ubyte value @A) clobbers(X) -> ubyte @A, ubyte @Y  {
 	; ---- A to hex petscii string in AY (first hex char in A, second hex char in Y)
 	%asm {{
-		stx  P8ZP_SCRATCH_REG
 		pha
 		and  #$0f
 		tax
@@ -714,7 +704,6 @@ asmsub  ubyte2hex  (ubyte value @A) -> ubyte @A, ubyte @Y  {
 		lsr  a
 		tax
 		lda  _hex_digits,x
-		ldx  P8ZP_SCRATCH_REG
 		rts
 
 _hex_digits	.text "0123456789abcdef"	; can probably be reused for other stuff as well

compiler/res/prog8lib/cx16/diskio.p8

@@ -6,6 +6,7 @@
 %import syslib
 
 diskio {
+    %option no_symbol_prefixing
 
     ubyte drivenumber = 8
 
@@ -42,12 +43,12 @@ diskio {
             ubyte high = cbm.CHRIN()
             txt.print_uw(mkword(high, low))
             txt.spc()
-            ubyte @zp char
+            ubyte @zp character
             repeat {
-                char = cbm.CHRIN()
-                if char==0
+                character = cbm.CHRIN()
+                if character==0
                     break
-                txt.chrout(char)
+                txt.chrout(character)
             }
             txt.nl()
             void cbm.CHRIN()     ; skip 2 bytes
@@ -73,39 +74,42 @@ io_error:
     }
 
     sub diskname() -> uword {
-        ; -- Returns pointer to disk name string or 0 if failure.
-
-        cbm.SETNAM(1, "$")
-        cbm.SETLFS(12, drivenumber, 0)
-        ubyte okay = false
-        void cbm.OPEN()          ; open 12,8,0,"$"
+        ; returns disk label name or 0 if error
+        cbm.SETNAM(3, "$")
+        cbm.SETLFS(12, diskio.drivenumber, 0)
+        ubyte status = 1
+        void cbm.OPEN()          ; open 12,8,0,"$=c"
         if_cs
             goto io_error
         void cbm.CHKIN(12)        ; use #12 as input channel
         if_cs
             goto io_error
 
-        repeat 6 {
-            void cbm.CHRIN()     ; skip the 6 prologue bytes
+        while cbm.CHRIN()!='"' {
+            ; skip up to entry name
         }
-        if cbm.READST()!=0
-            goto io_error
 
-        cx16.r0 = &list_filename
+        cx16.r0 = &diskio.list_filename
         repeat {
             @(cx16.r0) = cbm.CHRIN()
-            if @(cx16.r0)==0
+            if @(cx16.r0)=='"' {
+                @(cx16.r0) = ' '
+                while @(cx16.r0)==' ' and cx16.r0>=&diskio.list_filename {
+                    @(cx16.r0) = 0
+                    cx16.r0--
+                }
                 break
+            }
             cx16.r0++
         }
-        okay = true
+        status = cbm.READST()
 
 io_error:
-        cbm.CLRCHN()        ; restore default i/o devices
+        cbm.CLRCHN()
         cbm.CLOSE(12)
-        if okay
-            return &list_filename
-        return 0
+        if status and status & $40 == 0
+            return 0
+        return diskio.list_filename
     }
 
     ; internal variables for the iterative file lister / loader
@@ -205,12 +209,12 @@ io_error:
 
             ; read the filename
             repeat {
-                ubyte char = cbm.CHRIN()
-                if char==0
+                ubyte character = cbm.CHRIN()
+                if character==0
                     break
-                if char=='\"'
+                if character=='\"'
                     break
-                @(nameptr) = char
+                @(nameptr) = character
                 nameptr++
             }
 
@@ -286,25 +290,26 @@ close_end:
     sub f_read(uword bufferpointer, uword num_bytes) -> uword {
         ; -- read from the currently open file, up to the given number of bytes.
         ;    returns the actual number of bytes read.  (checks for End-of-file and error conditions)
+        ;    NOTE: cannot be used to load into VRAM.  Use vload() or call cx16.macptr() yourself with the vera data register as address.
         if not iteration_in_progress or not num_bytes
             return 0
 
         list_blocks = 0     ; we reuse this variable for the total number of bytes read
 
         ; commander X16 supports fast block-read via macptr() kernal call
-        uword size
+        uword readsize
         while num_bytes {
-            size = 255
-            if num_bytes<size
-                size = num_bytes
-            size = cx16.macptr(lsb(size), bufferpointer, false)
+            readsize = 255
+            if num_bytes<readsize
+                readsize = num_bytes
+            readsize = cx16.macptr(lsb(readsize), bufferpointer, false)
             if_cs
                 goto byte_read_loop     ; macptr block read not supported, do fallback loop
-            list_blocks += size
-            bufferpointer += size
+            list_blocks += readsize
+            bufferpointer += readsize
             if msb(bufferpointer) == $c0
                 bufferpointer = mkword($a0, lsb(bufferpointer))  ; wrap over bank boundary
-            num_bytes -= size
+            num_bytes -= readsize
             if cbm.READST() & $40 {
                 f_close()       ; end of file, close it
                 break
@@ -343,6 +348,7 @@ m_in_buffer     sta  $ffff
     ; optimized for Commander X16 to use MACPTR block read kernal call
     sub f_read_all(uword bufferpointer) -> uword {
         ; -- read the full contents of the file, returns number of bytes read.
+        ;    NOTE: cannot be used to load into VRAM.  Use vload() or call cx16.macptr() yourself with the vera data register as address.
         if not iteration_in_progress
             return 0
 
@@ -464,27 +470,26 @@ io_error:
 
 
     ; saves a block of memory to disk, including the default 2 byte prg header.
-    sub save(uword filenameptr, uword address, uword size) -> bool {
-        return internal_save_routine(filenameptr, address, size, false)
+    sub save(uword filenameptr, uword startaddress, uword savesize) -> bool {
+        return internal_save_routine(filenameptr, startaddress, savesize, false)
     }
 
     ; like save() but omits the 2 byte prg header.
-    sub save_raw(uword filenameptr, uword address, uword size) -> bool {
-        return internal_save_routine(filenameptr, address, size, true)
+    sub save_raw(uword filenameptr, uword startaddress, uword savesize) -> bool {
+        return internal_save_routine(filenameptr, startaddress, savesize, true)
     }
 
-    sub internal_save_routine(uword filenameptr, uword address, uword size, bool headerless) -> bool {
+    sub internal_save_routine(uword filenameptr, uword startaddress, uword savesize, bool headerless) -> bool {
         cbm.SETNAM(string.length(filenameptr), filenameptr)
         cbm.SETLFS(1, drivenumber, 0)
-        uword @shared end_address = address + size
+        uword @shared end_address = startaddress + savesize
         cx16.r0L = 0
 
         %asm {{
-            lda  address
+            lda  startaddress
             sta  P8ZP_SCRATCH_W1
-            lda  address+1
+            lda  startaddress+1
             sta  P8ZP_SCRATCH_W1+1
-            stx  P8ZP_SCRATCH_REG
             ldx  end_address
             ldy  end_address+1
             lda  headerless
@@ -495,7 +500,6 @@ io_error:
 +           lda  #<P8ZP_SCRATCH_W1
             jsr  cbm.SAVE
 +           php
-            ldx  P8ZP_SCRATCH_REG
             plp
         }}
 
@@ -528,8 +532,8 @@ io_error:
     ; Identical to load(), but DOES INCLUDE the first 2 bytes in the file.
     ; No program header is assumed in the file. Everything is loaded.
     ; See comments on load() for more details.
-    sub load_raw(uword filenameptr, uword address) -> uword {
-        return internal_load_routine(filenameptr, address, true)
+    sub load_raw(uword filenameptr, uword startaddress) -> uword {
+        return internal_load_routine(filenameptr, startaddress, true)
     }
 
 
@@ -543,7 +547,6 @@ io_error:
             secondary |= %00000010  ; activate cx16 kernal headerless load support
         cbm.SETLFS(1, drivenumber, secondary)
         %asm {{
-            stx  P8ZP_SCRATCH_REG
             lda  #0
             ldx  address_override
             ldy  address_override+1
@@ -551,7 +554,7 @@ io_error:
             bcs  +
             stx  cx16.r1
             sty  cx16.r1+1
-+           ldx  P8ZP_SCRATCH_REG
++
         }}
 
         cbm.CLRCHN()
@@ -603,14 +606,13 @@ io_error:
         return $2000 * (cx16.getrambank() - startbank) + endaddress - startaddress
     }
 
-    asmsub vload(str name @R0, ubyte bank @A, uword address @R1) -> ubyte @A {
+    asmsub vload(str name @R0, ubyte bank @A, uword startaddress @R1) clobbers(X, Y) -> ubyte @A {
         ; -- like the basic command VLOAD "filename",drivenumber,bank,address
         ;    loads a file into Vera's video memory in the given bank:address, returns success in A
         ;    the file has to have the usual 2 byte header (which will be skipped)
         %asm {{
             clc
 internal_vload:
-            phx
             pha
             ldx  drivenumber
             bcc +
@@ -638,13 +640,12 @@ internal_vload:
     +       jsr  cbm.CLRCHN
             lda  #1
             jsr  cbm.CLOSE
-            plx
             lda  P8ZP_SCRATCH_B1
             rts
         }}
     }
 
-    asmsub vload_raw(str name @R0, ubyte bank @A, uword address @R1) -> ubyte @A {
+    asmsub vload_raw(str name @R0, ubyte bank @A, uword startaddress @R1) clobbers(X, Y) -> ubyte @A {
         ; -- like the basic command BVLOAD "filename",drivenumber,bank,address
         ;    loads a file into Vera's video memory in the given bank:address, returns success in A
         ;    the file is read fully including the first two bytes.
@@ -684,6 +685,75 @@ internal_vload:
         send_command(list_filename)
     }
 
+    sub curdir() -> uword {
+        ; return current directory name or 0 if error
+        ; special X16 dos command to only return the current path in the entry list (R42+)
+        const ubyte MAX_PATH_LEN=80
+        uword reversebuffer = memory("curdir_buffer", MAX_PATH_LEN, 0)
+        cx16.r12 = reversebuffer + MAX_PATH_LEN-1
+        @(cx16.r12)=0
+        cbm.SETNAM(3, "$=c")
+        cbm.SETLFS(12, diskio.drivenumber, 0)
+        void cbm.OPEN()          ; open 12,8,0,"$=c"
+        if_cs
+            goto io_error
+        void cbm.CHKIN(12)        ; use #12 as input channel
+        if_cs
+            goto io_error
+
+        repeat 6 {
+            void cbm.CHRIN()
+        }
+        while cbm.CHRIN() {
+            ; skip first line (drive label)
+        }
+        while cbm.CHRIN()!='"' {
+            ; skip to first name
+        }
+        ubyte status = cbm.READST()
+        cx16.r10 = &list_filename
+        while status==0 {
+            repeat {
+                @(cx16.r10) = cbm.CHRIN()
+                if @(cx16.r10)==0
+                    break
+                cx16.r10++
+            }
+            while @(cx16.r10)!='"' and cx16.r10>=&list_filename {
+                @(cx16.r10)=0
+                cx16.r10--
+            }
+            @(cx16.r10)=0
+            prepend(list_filename)
+            cx16.r10 = &list_filename
+            while cbm.CHRIN()!='"' and status==0 {
+                status = cbm.READST()
+                ; skipping up to next entry name
+            }
+        }
+
+io_error:
+        cbm.CLRCHN()
+        cbm.CLOSE(12)
+        if status and status & $40 == 0
+            return 0
+        if @(cx16.r12)==0 {
+            cx16.r12--
+            @(cx16.r12)='/'
+        }
+        return cx16.r12
+
+        sub prepend(str dir) {
+            if dir[0]=='/' and dir[1]==0
+                return
+            cx16.r9L = string.length(dir)
+            cx16.r12 -= cx16.r9L
+            sys.memcopy(dir, cx16.r12, cx16.r9L)
+            cx16.r12--
+            @(cx16.r12)='/'
+        }
+    }
+
     sub relabel(str name) {
         ; -- change the disk label.
         list_filename[0] = 'r'

compiler/res/prog8lib/cx16/floats.p8

@@ -6,6 +6,8 @@
 floats {
 	; ---- this block contains C-64 compatible floating point related functions ----
 
+    %option no_symbol_prefixing
+
         const float  PI     = 3.141592653589793
         const float  TWOPI  = 6.283185307179586
 
@@ -26,7 +28,7 @@ romsub $fe00 = AYINT() clobbers(A,X,Y)          ; fac1-> signed word in 100-101
 romsub $fe03 = GIVAYF(ubyte lo @ Y, ubyte hi @ A) clobbers(A,X,Y)
 
 romsub $fe06 = FOUT() clobbers(X) -> uword @ AY             ; fac1 -> string, address returned in AY
-; romsub $fe09 = VAL_1() clobbers(A,X,Y)                      ; convert ASCII string to floating point [not yet implemented!!!]
+; romsub $fe09 = VAL_1() clobbers(A,X,Y)                      ; convert ASCII string to floating point [not yet implemented!!!] see parse_f() instead
 
 ; GETADR: fac1 -> unsigned word in Y/A (might throw ILLEGAL QUANTITY) (result also in $14/15)
 ; (tip: use GETADRAY to get A/Y output; lo/hi switched to normal little endian order)
@@ -42,7 +44,7 @@ romsub $fe21 = FMULTT() clobbers(A,X,Y)                     ; fac1 *= fac2
 romsub $fe24 = FDIV(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt in A/Y / fac1
 romsub $fe27 = FDIVT() clobbers(A,X,Y)                      ; fac1 = fac2/fac1  mind the order of the operands
 romsub $fe2a = LOG() clobbers(A,X,Y)                        ; fac1 = LN(fac1)  (natural log)
-romsub $fe2d = INT() clobbers(A,X,Y)                        ; INT() truncates, use FADDH first to round instead of trunc
+romsub $fe2d = INT() clobbers(A,X,Y)                        ; INT() truncates, use ROUND or FADDH first to round instead of trunc
 romsub $fe30 = SQR() clobbers(A,X,Y)                        ; fac1 = SQRT(fac1)
 romsub $fe33 = NEGOP() clobbers(A)                          ; switch the sign of fac1 (fac1 = -fac1)
 romsub $fe36 = FPWR(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = fac2 ** float in A/Y
@@ -97,14 +99,11 @@ asmsub  FREADSA  (byte value @A) clobbers(A,X,Y) {
 asmsub  GIVUAYFAY  (uword value @ AY) clobbers(A,X,Y)  {
 	; ---- unsigned 16 bit word in A/Y (lo/hi) to fac1
 	%asm {{
-	phx
 	sty  $c4        ; facmo     ($64 on c128)
 	sta  $c5        ; facmo+1   ($65 on c128)
 	ldx  #$90
 	sec
-	jsr  FLOATC
-	plx
-	rts
+	jmp  FLOATC
 	}}
 }
 
@@ -137,16 +136,26 @@ asmsub  FREADUY (ubyte value @Y) {
     }}
 }
 
+asmsub parse_f(str value @AY) -> float @FAC1 {
+    ; -- parse a string value of a number to float in FAC1
+    ;    warning: uses an internal BASIC routine that may be rom version dependent
+    ;    ($ddf2 is inside the routine for VAL at $ddef)
+    %asm {{
+        sta  $a9
+        sty  $aa
+        jsr  prog8_lib.strlen
+        tya
+        jmp  $ddf2
+    }}
+}
+
 
 &uword AYINT_facmo = $c6      ; $c6/$c7 contain result of AYINT
 
 sub rndf() -> float {
     %asm {{
-        phx
         lda  #1
-        jsr  RND_0
-        plx
-        rts
+        jmp  RND_0
     }}
 }
 

compiler/res/prog8lib/cx16/gfx2.p8

@@ -13,16 +13,17 @@
 ; SCREEN MODE LIST:
 ;   mode 0 = reset back to default text mode
 ;   mode 1 = bitmap 320 x 240 monochrome
-;   mode 2 = bitmap 320 x 240 x 4c (TODO not yet implemented)
-;   mode 3 = bitmap 320 x 240 x 16c (TODO not yet implemented)
+;   mode 2 = bitmap 320 x 240 x 4c (not yet implemented: just use 256c, there's enough vram for that)
+;   mode 3 = bitmap 320 x 240 x 16c (not yet implemented: just use 256c, there's enough vram for that)
 ;   mode 4 = bitmap 320 x 240 x 256c  (like SCREEN $80 but using this api instead of kernal)
 ;   mode 5 = bitmap 640 x 480 monochrome
 ;   mode 6 = bitmap 640 x 480 x 4c
 ;   higher color dephts in highres are not supported due to lack of VRAM
 
-
 gfx2 {
 
+    %option no_symbol_prefixing
+
     ; read-only control variables:
     ubyte active_mode = 0
     uword width = 0
@@ -44,7 +45,7 @@ gfx2 {
                 height = 240
                 bpp = 1
             }
-            ; TODO modes 2, 3 not yet implemented
+            ; TODO modes 2, 3
             4 -> {
                 ; lores 256c
                 cx16.VERA_DC_VIDEO = (cx16.VERA_DC_VIDEO & %11001111) | %00100000      ; enable only layer 1
@@ -107,7 +108,7 @@ gfx2 {
                 repeat 240/2/8
                     cs_innerloop640()
             }
-            ; TODO mode 2, 3
+            ; TODO modes 2, 3
             4 -> {
                 ; lores 256c
                 repeat 240/2
@@ -132,51 +133,61 @@ gfx2 {
         monochrome_dont_stipple_flag = not enable
     }
 
-    sub rect(uword x, uword y, uword rwidth, uword rheight, ubyte color) {
+    sub rect(uword xx, uword yy, uword rwidth, uword rheight, ubyte color) {
         if rwidth==0 or rheight==0
             return
-        horizontal_line(x, y, rwidth, color)
+        horizontal_line(xx, yy, rwidth, color)
         if rheight==1
             return
-        horizontal_line(x, y+rheight-1, rwidth, color)
-        vertical_line(x, y+1, rheight-2, color)
+        horizontal_line(xx, yy+rheight-1, rwidth, color)
+        vertical_line(xx, yy+1, rheight-2, color)
         if rwidth==1
             return
-        vertical_line(x+rwidth-1, y+1, rheight-2, color)
+        vertical_line(xx+rwidth-1, yy+1, rheight-2, color)
     }
 
-    sub fillrect(uword x, uword y, uword rwidth, uword rheight, ubyte color) {
+    sub fillrect(uword xx, uword yy, uword rwidth, uword rheight, ubyte color) {
         if rwidth==0
             return
         repeat rheight {
-            horizontal_line(x, y, rwidth, color)
-            y++
+            horizontal_line(xx, yy, rwidth, color)
+            yy++
         }
     }
 
-    sub horizontal_line(uword x, uword y, uword length, ubyte color) {
-        ubyte[9] masked_ends   = [ 0, %10000000, %11000000, %11100000, %11110000, %11111000, %11111100, %11111110, %11111111]
+    sub horizontal_line(uword xx, uword yy, uword length, ubyte color) {
         ubyte[9] masked_starts = [ 0, %00000001, %00000011, %00000111, %00001111, %00011111, %00111111, %01111111, %11111111]
+        ubyte[9] masked_ends   = [ 0, %10000000, %11000000, %11100000, %11110000, %11111000, %11111100, %11111110, %11111111]
 
         if length==0
             return
         when active_mode {
             1, 5 -> {
                 ; monochrome modes, either resolution
-                ubyte separate_pixels = (8-lsb(x)) & 7
+                ubyte separate_pixels = (8-lsb(xx)) & 7
                 if separate_pixels as uword > length
                     separate_pixels = lsb(length)
                 if separate_pixels {
-                    position(x,y)
-                    cx16.VERA_ADDR_H &= %00000111   ; vera auto-increment off
-                    cx16.VERA_DATA0 = cx16.VERA_DATA0 | masked_starts[separate_pixels]
+                    if monochrome_dont_stipple_flag {
+                        position(xx,yy)
+                        cx16.VERA_ADDR_H &= %00000111   ; vera auto-increment off
+                        if color
+                            cx16.VERA_DATA0 |= masked_starts[separate_pixels]
+                        else
+                            cx16.VERA_DATA0 &= ~masked_starts[separate_pixels]
+                        xx += separate_pixels
+                    } else {
+                        repeat separate_pixels {
+                            plot(xx, yy, color)
+                            xx++
+                        }
+                    }
                     length -= separate_pixels
-                    x += separate_pixels
                 }
                 if length {
-                    position(x, y)
+                    position(xx, yy)
                     separate_pixels = lsb(length) & 7
-                    x += length & $fff8
+                    xx += length & $fff8
                     %asm {{
                         lsr  length+1
                         ror  length
@@ -192,7 +203,7 @@ gfx2 {
                         beq  _stipple
                         ldy  #255       ; don't stipple
                         bra  _loop
-_stipple                lda  y
+_stipple                lda  yy
                         and  #1         ; determine stipple pattern to use
                         bne  +
                         ldy  #%01010101
@@ -210,17 +221,26 @@ _loop                   lda  length
 _done
                     }}
 
-                    cx16.VERA_ADDR_H &= %00000111   ; vera auto-increment off
-                    cx16.VERA_DATA0 = cx16.VERA_DATA0 | masked_ends[separate_pixels]
+                    if monochrome_dont_stipple_flag {
+                        cx16.VERA_ADDR_H &= %00000111   ; vera auto-increment off
+                        if color
+                            cx16.VERA_DATA0 |= masked_ends[separate_pixels]
+                        else
+                            cx16.VERA_DATA0 &= ~masked_ends[separate_pixels]
+                    } else {
+                        repeat separate_pixels {
+                            plot(xx, yy, color)
+                            xx++
+                        }
+                    }
                 }
                 cx16.VERA_ADDR_H &= %00000111   ; vera auto-increment off again
             }
             4 -> {
                 ; lores 256c
-                position(x, y)
+                position(xx, yy)
                 %asm {{
                     lda  color
-                    phx
                     ldx  length+1
                     beq  +
                     ldy  #0
@@ -234,12 +254,11 @@ _done
 -                   sta  cx16.VERA_DATA0
                     dey
                     bne  -
-+                   plx
++
                 }}
             }
             6 -> {
-                ; highres 4c
-                ; TODO also mostly usable for lores 4c?
+                ; highres 4c ....also mostly usable for mode 2, lores 4c?
                 color &= 3
                 ubyte[4] colorbits
                 ubyte ii
@@ -247,7 +266,7 @@ _done
                     colorbits[ii] = color
                     color <<= 2
                 }
-                void addr_mul_24_for_highres_4c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+                void addr_mul_24_for_highres_4c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
                 %asm {{
                     lda  cx16.VERA_ADDR_H
                     and  #%00000111         ; no auto advance
@@ -260,8 +279,7 @@ _done
                     sta  cx16.VERA_ADDR_L
                     lda  cx16.r0+1
                     sta  cx16.VERA_ADDR_M
-                    phx
-                    ldx  x
+                    ldx  xx
                 }}
 
                 repeat length {
@@ -283,23 +301,19 @@ _done
 +                       inx                     ; next pixel
                     }}
                 }
-
-                %asm {{
-                    plx
-                }}
             }
         }
     }
 
-    sub vertical_line(uword x, uword y, uword lheight, ubyte color) {
+    sub vertical_line(uword xx, uword yy, uword lheight, ubyte color) {
         when active_mode {
             1, 5 -> {
                 ; monochrome, lo-res
-                cx16.r15L = gfx2.plot.bits[x as ubyte & 7]           ; bitmask
+                cx16.r15L = gfx2.plot.maskbits[xx as ubyte & 7]           ; bitmask
                 if color {
                     if monochrome_dont_stipple_flag {
                         ; draw continuous line.
-                        position2(x,y,true)
+                        position2(xx,yy,true)
                         if active_mode==1
                             set_both_strides(11)    ; 40 increment = 1 line in 320 px monochrome
                         else
@@ -313,11 +327,11 @@ _done
                         }
                     } else {
                         ; draw stippled line.
-                        if x&1 {
-                            y++
+                        if xx&1 {
+                            yy++
                             lheight--
                         }
-                        position2(x,y,true)
+                        position2(xx,yy,true)
                         if active_mode==1
                             set_both_strides(12)    ; 80 increment = 2 line in 320 px monochrome
                         else
@@ -331,7 +345,7 @@ _done
                         }
                     }
                 } else {
-                    position2(x,y,true)
+                    position2(xx,yy,true)
                     cx16.r15 = ~cx16.r15    ; erase pixels
                     if active_mode==1
                         set_both_strides(11)    ; 40 increment = 1 line in 320 px monochrome
@@ -349,7 +363,7 @@ _done
             4 -> {
                 ; lores 256c
                 ; set vera auto-increment to 320 pixel increment (=next line)
-                position(x,y)
+                position(xx,yy)
                 cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | (14<<4)
                 %asm {{
                     ldy  lheight
@@ -366,17 +380,17 @@ _done
                 ; use TWO vera adress pointers simultaneously one for reading, one for writing, so auto-increment is possible
                 if lheight==0
                     return
-                position2(x,y,true)
+                position2(xx,yy,true)
                 set_both_strides(13)    ; 160 increment = 1 line in 640 px 4c mode
                 ;; color &= 3
-                ;; color <<= gfx2.plot.shift4c[lsb(x) & 3]
-                cx16.r2L = lsb(x) & 3
+                ;; color <<= gfx2.plot.shift4c[lsb(xx) & 3]
+                cx16.r2L = lsb(xx) & 3
                 when color & 3 {
                     1 -> color = gfx2.plot.shiftedleft_4c_1[cx16.r2L]
                     2 -> color = gfx2.plot.shiftedleft_4c_2[cx16.r2L]
                     3 -> color = gfx2.plot.shiftedleft_4c_3[cx16.r2L]
                 }
-                ubyte @shared mask = gfx2.plot.mask4c[lsb(x) & 3]
+                ubyte @shared mask = gfx2.plot.mask4c[lsb(xx) & 3]
                 repeat lheight {
                     %asm {{
                         lda  cx16.VERA_DATA0
@@ -425,17 +439,17 @@ _done
         }
 
         word @zp d = 0
-        cx16.r13 = true      ; 'positive_ix'
+        cx16.r1L = true      ; 'positive_ix'
         if dx < 0 {
             dx = -dx
-            cx16.r13 = false
+            cx16.r1L = false
         }
         word @zp dx2 = dx*2
         word @zp dy2 = dy*2
         cx16.r14 = x1       ; internal plot X
 
         if dx >= dy {
-            if cx16.r13 {
+            if cx16.r1L {
                 repeat {
                     plot(cx16.r14, y1, color)
                     if cx16.r14==x2
@@ -462,7 +476,7 @@ _done
             }
         }
         else {
-            if cx16.r13 {
+            if cx16.r1L {
                 repeat {
                     plot(cx16.r14, y1, color)
                     if y1 == y2
@@ -555,8 +569,8 @@ _done
         }
     }
 
-    sub plot(uword @zp x, uword @zp y, ubyte @zp color) {
-        ubyte[8] @shared bits = [128, 64, 32, 16, 8, 4, 2, 1]
+    sub plot(uword @zp xx, uword @zp yy, ubyte @zp color) {
+        ubyte[8] @shared maskbits = [128, 64, 32, 16, 8, 4, 2, 1]
         ubyte[4] @shared mask4c = [%00111111, %11001111, %11110011, %11111100]
         ubyte[4] @shared shift4c = [6,4,2,0]
         ubyte[4] shiftedleft_4c_1 = [1<<6, 1<<4, 1<<2, 1<<0]
@@ -566,42 +580,32 @@ _done
         when active_mode {
             1 -> {
                 ; lores monochrome
-                %asm {{
-                    lda  x
-                    eor  y
-                    ora  monochrome_dont_stipple_flag
-                    and  #1
-                }}
-                if_nz {
+                if color {
+                    ; solid color or perhaps stipple
                     %asm {{
-                        lda  x
-                        and  #7
-                        pha     ; xbits
+                        lda  xx
+                        eor  yy
+                        ora  monochrome_dont_stipple_flag
+                        and  #1
                     }}
-                    x /= 8
-                    x += y*(320/8)
+                    if_nz {
+                        mode_1_prepare()
+                        %asm {{
+                            tsb  cx16.VERA_DATA0
+                        }}
+                    }
+                } else {
+                    ; only erase
+                    mode_1_prepare()
                     %asm {{
-                        stz  cx16.VERA_CTRL
-                        stz  cx16.VERA_ADDR_H
-                        lda  x+1
-                        sta  cx16.VERA_ADDR_M
-                        lda  x
-                        sta  cx16.VERA_ADDR_L
-                        ply         ; xbits
-                        lda  bits,y
-                        ldy  color
-                        beq  +
-                        tsb  cx16.VERA_DATA0
-                        bra  ++
-+                       trb  cx16.VERA_DATA0
-+
+                        trb  cx16.VERA_DATA0
                     }}
                 }
             }
-            ; TODO mode 2,3
+            ; TODO modes 2, 3
             4 -> {
                 ; lores 256c
-                void addr_mul_24_for_lores_256c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+                void addr_mul_24_for_lores_256c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
                 %asm {{
                     stz  cx16.VERA_CTRL
                     lda  cx16.r1
@@ -617,43 +621,32 @@ _done
             }
             5 -> {
                 ; highres monochrome
-                %asm {{
-                    lda  x
-                    eor  y
-                    ora  monochrome_dont_stipple_flag
-                    and  #1
-                }}
-                if_nz {
+                if color {
+                    ; solid color or perhaps stipple
                     %asm {{
-                        lda  x
-                        and  #7
-                        pha     ; xbits
+                        lda  xx
+                        eor  yy
+                        ora  monochrome_dont_stipple_flag
+                        and  #1
                     }}
-                    x /= 8
-                    x += y*(640/8)
+                    if_nz {
+                        mode_5_prepare()
+                        %asm {{
+                            tsb  cx16.VERA_DATA0
+                        }}
+                    }
+                } else {
+                    ; only erase
+                    mode_5_prepare()
                     %asm {{
-                        stz  cx16.VERA_CTRL
-                        stz  cx16.VERA_ADDR_H
-                        lda  x+1
-                        sta  cx16.VERA_ADDR_M
-                        lda  x
-                        sta  cx16.VERA_ADDR_L
-                        ply     ; xbits
-                        lda  bits,y
-                        ldy  color
-                        beq  +
-                        tsb  cx16.VERA_DATA0
-                        bra  ++
-+                       trb  cx16.VERA_DATA0
-+
+                        trb  cx16.VERA_DATA0
                     }}
                 }
             }
             6 -> {
-                ; highres 4c
-                ; TODO also mostly usable for lores 4c?
-                void addr_mul_24_for_highres_4c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
-                cx16.r2L = lsb(x) & 3       ; xbits
+                ; highres 4c   ....also mostly usable for mode 2, lores 4c?
+                void addr_mul_24_for_highres_4c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
+                cx16.r2L = lsb(xx) & 3       ; xbits
                 ; color &= 3
                 ; color <<= shift4c[cx16.r2L]
                 when color & 3 {
@@ -677,38 +670,78 @@ _done
                 }}
             }
         }
+
+        sub mode_1_prepare() {
+            %asm {{
+                lda  xx
+                and  #7
+                pha     ; xbits
+            }}
+            xx /= 8
+            xx += yy*(320/8)
+            %asm {{
+                stz  cx16.VERA_CTRL
+                stz  cx16.VERA_ADDR_H
+                lda  xx+1
+                sta  cx16.VERA_ADDR_M
+                lda  xx
+                sta  cx16.VERA_ADDR_L
+                ply     ; xbits
+                lda  maskbits,y
+            }}
+        }
+
+        sub mode_5_prepare() {
+            %asm {{
+                lda  xx
+                and  #7
+                pha     ; xbits
+            }}
+            xx /= 8
+            xx += yy*(640/8)
+            %asm {{
+                stz  cx16.VERA_CTRL
+                stz  cx16.VERA_ADDR_H
+                lda  xx+1
+                sta  cx16.VERA_ADDR_M
+                lda  xx
+                sta  cx16.VERA_ADDR_L
+                ply     ; xbits
+                lda  maskbits,y
+            }}
+        }
     }
 
-    sub pget(uword @zp x, uword y) -> ubyte {
+    sub pget(uword @zp xx, uword yy) -> ubyte {
         when active_mode {
             1 -> {
                 ; lores monochrome
                 %asm {{
-                    lda  x
+                    lda  xx
                     and  #7
                     pha     ; xbits
                 }}
-                x /= 8
-                x += y*(320/8)
+                xx /= 8
+                xx += yy*(320/8)
                 %asm {{
                     stz  cx16.VERA_CTRL
                     stz  cx16.VERA_ADDR_H
-                    lda  x+1
+                    lda  xx+1
                     sta  cx16.VERA_ADDR_M
-                    lda  x
+                    lda  xx
                     sta  cx16.VERA_ADDR_L
                     ply         ; xbits
-                    lda  plot.bits,y
+                    lda  plot.maskbits,y
                     and  cx16.VERA_DATA0
                     beq  +
                     lda  #1
 +
                 }}
             }
-            ; TODO mode 2 and 3
+            ; TODO modes 2, 3
             4 -> {
                 ; lores 256c
-                void addr_mul_24_for_lores_256c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+                void addr_mul_24_for_lores_256c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
                 %asm {{
                     stz  cx16.VERA_CTRL
                     lda  cx16.r1
@@ -723,21 +756,21 @@ _done
             5 -> {
                 ; hires monochrome
                 %asm {{
-                    lda  x
+                    lda  xx
                     and  #7
                     pha     ; xbits
                 }}
-                x /= 8
-                x += y*(640/8)
+                xx /= 8
+                xx += yy*(640/8)
                 %asm {{
                     stz  cx16.VERA_CTRL
                     stz  cx16.VERA_ADDR_H
-                    lda  x+1
+                    lda  xx+1
                     sta  cx16.VERA_ADDR_M
-                    lda  x
+                    lda  xx
                     sta  cx16.VERA_ADDR_L
                     ply     ; xbits
-                    lda  plot.bits,y
+                    lda  plot.maskbits,y
                     and  cx16.VERA_DATA0
                     beq  +
                     lda  #1
@@ -746,7 +779,7 @@ _done
             }
             6 -> {
                 ; hires 4c
-                void addr_mul_24_for_highres_4c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+                void addr_mul_24_for_highres_4c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
                 %asm {{
                     stz  cx16.VERA_CTRL
                     lda  cx16.r1L
@@ -758,7 +791,7 @@ _done
                     lda  cx16.VERA_DATA0
                     sta  cx16.r0L
                 }}
-                cx16.r1L = lsb(x) & 3
+                cx16.r1L = lsb(xx) & 3
                 cx16.r0L >>= gfx2.plot.shift4c[cx16.r1L]
                 return cx16.r0L & 3
             }
@@ -766,7 +799,7 @@ _done
         }
     }
 
-    sub fill(word @zp x, word @zp y, ubyte new_color) {
+    sub fill(word @zp xx, word @zp yy, ubyte new_color) {
         ; Non-recursive scanline flood fill.
         ; based loosely on code found here https://www.codeproject.com/Articles/6017/QuickFill-An-efficient-flood-fill-algorithm
         ; with the fixes applied to the seedfill_4 routine as mentioned in the comments.
@@ -829,96 +862,87 @@ _done
                 lda  stack_xr_msb,y
                 sta  x2+1
                 lda  stack_y_lsb,y
-                sta  y
+                sta  yy
                 lda  stack_y_msb,y
-                sta  y+1
+                sta  yy+1
                 ldy  cx16.r12L
                 lda  stack_dy,y
                 sta  dy
             }}
-            y+=dy
+            yy+=dy
         }
-        cx16.r11L = pget(x as uword, y as uword)        ; old_color
+        cx16.r11L = pget(xx as uword, yy as uword)        ; old_color
         if cx16.r11L == cx16.r10L
             return
-        if x<0 or x > width-1 or y<0 or y > height-1
+        if xx<0 or xx>width-1 or yy<0 or yy>height-1
             return
-        push_stack(x, x, y, 1)
-        push_stack(x, x, y + 1, -1)
+        push_stack(xx, xx, yy, 1)
+        push_stack(xx, xx, yy + 1, -1)
         word left = 0
         while cx16.r12L {
             pop_stack()
-            x = x1
-            while x >= 0 and pget(x as uword, y as uword) == cx16.r11L {
-                plot(x as uword, y as uword, cx16.r10L)
-                x--
+            xx = x1
+            while xx >= 0 and pget(xx as uword, yy as uword) == cx16.r11L {
+                plot(xx as uword, yy as uword, cx16.r10L)
+                xx--
             }
-            if x>= x1
+            if xx >= x1
                 goto skip
 
-            left = x + 1
+            left = xx + 1
             if left < x1
-                push_stack(left, x1 - 1, y, -dy)
-            x = x1 + 1
+                push_stack(left, x1 - 1, yy, -dy)
+            xx = x1 + 1
 
             do {
-                while x <= width-1 and pget(x as uword, y as uword) == cx16.r11L {
-                    plot(x as uword, y as uword, cx16.r10L)
-                    x++
+                while xx <= width-1 and pget(xx as uword, yy as uword) == cx16.r11L {
+                    plot(xx as uword, yy as uword, cx16.r10L)
+                    xx++
                 }
-                push_stack(left, x - 1, y, dy)
-                if x > x2 + 1
-                    push_stack(x2 + 1, x - 1, y, -dy)
+                push_stack(left, xx - 1, yy, dy)
+                if xx > x2 + 1
+                    push_stack(x2 + 1, xx - 1, yy, -dy)
 skip:
-                x++
-                while x <= x2 and pget(x as uword, y as uword) != cx16.r11L
-                    x++
-                left = x
-            } until x>x2
+                xx++
+                while xx <= x2 and pget(xx as uword, yy as uword) != cx16.r11L
+                    xx++
+                left = xx
+            } until xx>x2
         }
     }
 
-    sub position(uword @zp x, uword y) {
-        ubyte bank
+    sub position(uword @zp xx, uword yy) {
         when active_mode {
             1 -> {
                 ; lores monochrome
-                cx16.r0 = y*(320/8) + x/8
+                cx16.r0 = yy*(320/8) + xx/8
                 cx16.vaddr(0, cx16.r0, 0, 1)
             }
-            ; TODO modes 2,3
+            ; TODO modes 2, 3
             4 -> {
                 ; lores 256c
-                void addr_mul_24_for_lores_256c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
-                bank = lsb(cx16.r1)
-                cx16.vaddr(bank, cx16.r0, 0, 1)
+                void addr_mul_24_for_lores_256c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
+                cx16.r2L = cx16.r1L
+                cx16.vaddr(cx16.r2L, cx16.r0, 0, 1)
             }
             5 -> {
                 ; highres monochrome
-                cx16.r0 = y*(640/8) + x/8
+                cx16.r0 = yy*(640/8) + xx/8
                 cx16.vaddr(0, cx16.r0, 0, 1)
             }
             6 -> {
                 ; highres 4c
-                void addr_mul_24_for_highres_4c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
-                bank = lsb(cx16.r1)
-                cx16.vaddr(bank, cx16.r0, 0, 1)
+                void addr_mul_24_for_highres_4c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
+                cx16.r2L = cx16.r1L
+                cx16.vaddr(cx16.r2L, cx16.r0, 0, 1)
             }
         }
     }
 
-    sub position2(uword @zp x, uword y, bool also_port_1) {
-        position(x, y)
-        if also_port_1 {
-            when active_mode {
-                1, 5 -> cx16.vaddr(0, cx16.r0, 1, 1)
-                ; TODO modes 2, 3
-                4, 6 -> {
-                    ubyte bank = lsb(cx16.r1)
-                    cx16.vaddr(bank, cx16.r0, 1, 1)
-                }
-            }
-        }
+    sub position2(uword @zp xx, uword yy, bool also_port_1) {
+        position(xx, yy)
+        if also_port_1
+            cx16.vaddr_clone(0)
     }
 
     inline asmsub next_pixel(ubyte color @A) {
@@ -931,13 +955,12 @@ skip:
         }}
     }
 
-    asmsub next_pixels(uword pixels @AY, uword amount @R0) clobbers(A, Y)  {
+    asmsub next_pixels(uword pixels @AY, uword amount @R0) clobbers(A, X, Y)  {
         ; -- sets the next bunch of pixels from a prepared array of bytes.
         ;    for 8 bpp screens this will plot 1 pixel per byte.
         ;    for 1 bpp screens it will plot 8 pixels at once (colors are the bit patterns per byte).
         ;    for 2 bpp screens it will plot 4 pixels at once (colors are the bit patterns per byte).
         %asm {{
-            phx
             sta  P8ZP_SCRATCH_W1
             sty  P8ZP_SCRATCH_W1+1
             ldx  cx16.r0+1
@@ -959,14 +982,13 @@ skip:
             iny
             dex
             bne  -
-+           plx
++           rts
         }}
     }
 
-    asmsub set_8_pixels_from_bits(ubyte bits @R0, ubyte oncolor @A, ubyte offcolor @Y) {
+    asmsub set_8_pixels_from_bits(ubyte bits @R0, ubyte oncolor @A, ubyte offcolor @Y) clobbers(X) {
         ; this is only useful in 256 color mode where one pixel equals one byte value.
         %asm {{
-            phx
             ldx  #8
 -           asl  cx16.r0
             bcc  +
@@ -975,7 +997,6 @@ skip:
 +           sty  cx16.VERA_DATA0
 +           dex
             bne  -
-            plx
             rts
         }}
     }
@@ -990,51 +1011,97 @@ skip:
         cx16.screen_set_charset(charset, 0)
     }
 
-    sub text(uword @zp x, uword y, ubyte color, uword sctextptr) {
+    sub text(uword @zp xx, uword yy, ubyte color, uword sctextptr) {
         ; -- Write some text at the given pixel position. The text string must be in screencode encoding (not petscii!).
         ;    You must also have called text_charset() first to select and prepare the character set to use.
-        ;    NOTE: in monochrome (1bpp) screen modes, x position is currently constrained to multiples of 8 !  TODO allow per-pixel horizontal positioning
-        ; TODO draw whole horizontal spans using vera auto increment if possible, instead of per-character columns
         uword chardataptr
+        ubyte[8] @shared char_bitmap_bytes_left
+        ubyte[8] @shared char_bitmap_bytes_right
+
         when active_mode {
             1, 5 -> {
                 ; monochrome mode, either resolution
-                cx16.r2 = 40
-                if active_mode==5
-                    cx16.r2 = 80
-                while @(sctextptr) {
-                    chardataptr = charset_addr + (@(sctextptr) as uword)*8
-                    cx16.vaddr(charset_bank, chardataptr, 1, 1)
-                    position(x,y)
+                cx16.r3 = sctextptr
+                while @(cx16.r3) {
+                    chardataptr = charset_addr + @(cx16.r3) * $0008
+                    ; copy the character bitmap into RAM
+                    cx16.vaddr_autoincr(charset_bank, chardataptr, 0, 1)
                     %asm {{
-                        lda  cx16.VERA_ADDR_H
-                        and  #%111              ; don't auto-increment, we have to do that manually because of the ora
-                        sta  cx16.VERA_ADDR_H
-                        lda  color
+                        ; pre-shift the bits
+                        lda  text.xx
+                        and  #7
                         sta  P8ZP_SCRATCH_B1
-                        ldy  #8
--                       lda  P8ZP_SCRATCH_B1
-                        bne  +                  ; white color, plot normally
-                        lda  cx16.VERA_DATA1
-                        eor  #255               ; black color, keep only the other pixels
-                        and  cx16.VERA_DATA0
-                        bra  ++
-+                       lda  cx16.VERA_DATA0
-                        ora  cx16.VERA_DATA1
-+                       sta  cx16.VERA_DATA0
-                        lda  cx16.VERA_ADDR_L
-                        clc
-                        adc  cx16.r2
-                        sta  cx16.VERA_ADDR_L
-                        bcc  +
-                        inc  cx16.VERA_ADDR_M
-+                       inc  x
-                        bne  +
-                        inc  x+1
-+                       dey
+                        ldy  #0
+-                       lda  cx16.VERA_DATA0
+                        stz  P8ZP_SCRATCH_REG
+                        ldx  P8ZP_SCRATCH_B1
+                        cpx  #0
+                        beq  +
+-                       lsr  a
+                        ror  P8ZP_SCRATCH_REG
+                        dex
                         bne  -
++                       sta  char_bitmap_bytes_left,y
+                        lda  P8ZP_SCRATCH_REG
+                        sta  char_bitmap_bytes_right,y
+                        iny
+                        cpy  #8
+                        bne  --
                     }}
-                    sctextptr++
+                    ; left part of shifted char
+                    position2(xx, yy, true)
+                    set_autoincrs_mode1_or_5()
+                    if color {
+                        %asm {{
+                            ldy  #0
+-                           lda  char_bitmap_bytes_left,y
+                            ora  cx16.VERA_DATA1
+                            sta  cx16.VERA_DATA0
+                            iny
+                            cpy  #8
+                            bne  -
+                        }}
+                    } else {
+                        %asm {{
+                            ldy  #0
+-                           lda  char_bitmap_bytes_left,y
+                            eor  #255
+                            and  cx16.VERA_DATA1
+                            sta  cx16.VERA_DATA0
+                            iny
+                            cpy  #8
+                            bne  -
+                        }}
+                    }
+                    ; right part of shifted char
+                    if lsb(xx) & 7 {
+                        position2(xx+8, yy, true)
+                        set_autoincrs_mode1_or_5()
+                        if color {
+                            %asm {{
+                                ldy  #0
+    -                           lda  char_bitmap_bytes_right,y
+                                ora  cx16.VERA_DATA1
+                                sta  cx16.VERA_DATA0
+                                iny
+                                cpy  #8
+                                bne  -
+                            }}
+                        } else {
+                            %asm {{
+                                ldy  #0
+    -                           lda  char_bitmap_bytes_right,y
+                                eor  #255
+                                and  cx16.VERA_DATA1
+                                sta  cx16.VERA_DATA0
+                                iny
+                                cpy  #8
+                                bne  -
+                            }}
+                        }
+                    }
+                    cx16.r3++
+                    xx += 8
                 }
             }
             4 -> {
@@ -1043,10 +1110,9 @@ skip:
                     chardataptr = charset_addr + (@(sctextptr) as uword)*8
                     cx16.vaddr(charset_bank, chardataptr, 1, 1)
                     repeat 8 {
-                        position(x,y)
-                        y++
+                        position(xx,yy)
+                        yy++
                         %asm {{
-                            phx
                             ldx  color
                             lda  cx16.VERA_DATA1
                             sta  P8ZP_SCRATCH_B1
@@ -1058,11 +1124,10 @@ skip:
 +                           lda  cx16.VERA_DATA0    ; don't write a pixel, but do advance to the next address
 +                           dey
                             bne  -
-                            plx
                         }}
                     }
-                    x+=8
-                    y-=8
+                    xx+=8
+                    yy-=8
                     sctextptr++
                 }
             }
@@ -1070,30 +1135,80 @@ skip:
                 ; hires 4c
                 ; we're going to use a few cx16 registers to make sure every variable is in zeropage in the inner loop.
                 cx16.r11L = color
-                cx16.r8 = y
                 while @(sctextptr) {
                     chardataptr = charset_addr + (@(sctextptr) as uword)*8
+                    cx16.vaddr(charset_bank, chardataptr, 1, true)  ; for reading the chardata from Vera data channel 1
+                    position(xx, yy)              ; only calculated once, we update vera address in the loop instead
+                    cx16.VERA_ADDR_H &= $0f     ; no auto increment
                     repeat 8 {
-                        ; TODO rewrite this inner loop partly in assembly
-                        ;      requires expanding the charbits to 2-bits per pixel (based on color)
-                        ;      also it's way more efficient to draw whole horizontal spans instead of per-character
-                        cx16.r9L = cx16.vpeek(charset_bank, chardataptr)  ; get the 8 horizontal character bits
-                        cx16.r7 = x
+                        cx16.r10L = cx16.VERA_DATA1  ; get the next 8 horizontal character bits
+                        cx16.r7 = xx
                         repeat 8 {
-                            cx16.r9L <<= 1
-                            if_cs
-                                plot(cx16.r7, cx16.r8, cx16.r11L)
+                            cx16.r10L <<= 1
+                            if_cs {
+                                cx16.r2L = cx16.r7L & 3       ; xbits
+                                when cx16.r11L & 3 {
+                                    1 -> cx16.r12L = gfx2.plot.shiftedleft_4c_1[cx16.r2L]
+                                    2 -> cx16.r12L = gfx2.plot.shiftedleft_4c_2[cx16.r2L]
+                                    3 -> cx16.r12L = gfx2.plot.shiftedleft_4c_3[cx16.r2L]
+                                    else -> cx16.r12L = 0
+                                }
+                                cx16.VERA_DATA0 = cx16.VERA_DATA0 & gfx2.plot.mask4c[cx16.r2L] | cx16.r12L
+                            }
                             cx16.r7++
+                            if (cx16.r7 & 3) == 0 {
+                                ; increment the pixel address by one
+                                %asm {{
+                                    stz  cx16.VERA_CTRL
+                                    clc
+                                    lda  cx16.VERA_ADDR_L
+                                    adc  #1
+                                    sta  cx16.VERA_ADDR_L
+                                    lda  cx16.VERA_ADDR_M
+                                    adc  #0
+                                    sta  cx16.VERA_ADDR_M
+                                    lda  cx16.VERA_ADDR_H
+                                    adc  #0
+                                    sta  cx16.VERA_ADDR_H
+                                }}
+                            }
                         }
-                        chardataptr++
-                        cx16.r8++
+
+                        ; increment pixel address to the next line
+                        %asm {{
+                            stz  cx16.VERA_CTRL
+                            clc
+                            lda  cx16.VERA_ADDR_L
+                            adc  #(640-8)/4
+                            sta  cx16.VERA_ADDR_L
+                            lda  cx16.VERA_ADDR_M
+                            adc  #0
+                            sta  cx16.VERA_ADDR_M
+                            lda  cx16.VERA_ADDR_H
+                            adc  #0
+                            sta  cx16.VERA_ADDR_H
+                        }}
                     }
-                    x+=8
-                    cx16.r8-=8
+                    xx+=8
                     sctextptr++
                 }
             }
         }
+
+        sub set_autoincrs_mode1_or_5() {
+            ; set autoincrements to go to next pixel row (40 or 80 increment)
+            if active_mode==1 {
+                cx16.VERA_CTRL = 0
+                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & $0f | (11<<4)
+                cx16.VERA_CTRL = 1
+                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & $0f | (11<<4)
+            } else {
+                cx16.VERA_CTRL = 0
+                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & $0f | (12<<4)
+                cx16.VERA_CTRL = 1
+                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & $0f | (12<<4)
+            }
+        }
     }
 
     asmsub cs_innerloop640() clobbers(Y) {

compiler/res/prog8lib/cx16/graphics.p8

@@ -10,6 +10,8 @@
 
 
 graphics {
+    %option no_symbol_prefixing
+
     const uword WIDTH = 320
     const ubyte HEIGHT = 240
 
@@ -27,7 +29,7 @@ graphics {
     sub disable_bitmap_mode() {
         ; enables text mode, erase the text screen, color white
         void cx16.screen_mode(0, false)
-        txt.fill_screen(' ', 1)     ; doesn't seem to fully clear the text screen after returning from gfx mode
+        txt.clear_screen()
     }
 
 
@@ -48,22 +50,22 @@ graphics {
         cx16.GRAPH_draw_line(x1, y1, x2, y2)
     }
 
-    sub fillrect(uword x, uword y, uword width, uword height) {
-        cx16.GRAPH_draw_rect(x, y, width, height, 0, 1)
+    sub fillrect(uword xx, uword yy, uword width, uword height) {
+        cx16.GRAPH_draw_rect(xx, yy, width, height, 0, 1)
     }
 
-    sub rect(uword x, uword y, uword width, uword height) {
-        cx16.GRAPH_draw_rect(x, y, width, height, 0, 0)
+    sub rect(uword xx, uword yy, uword width, uword height) {
+        cx16.GRAPH_draw_rect(xx, yy, width, height, 0, 0)
     }
 
-    sub horizontal_line(uword x, uword y, uword length) {
+    sub horizontal_line(uword xx, uword yy, uword length) {
         if length
-            cx16.GRAPH_draw_line(x, y, x+length-1, y)
+            cx16.GRAPH_draw_line(xx, yy, xx+length-1, yy)
     }
 
-    sub vertical_line(uword x, uword y, uword height) {
+    sub vertical_line(uword xx, uword yy, uword height) {
         if height
-            cx16.GRAPH_draw_line(x, y, x, y+height-1)
+            cx16.GRAPH_draw_line(xx, yy, xx, yy+height-1)
     }
 
     sub circle(uword xcenter, ubyte ycenter, ubyte radius) {

compiler/res/prog8lib/cx16/palette.p8

@@ -2,9 +2,10 @@
 ; Should you want to restore the default palette, you have to reinitialize the Vera yourself.
 
 palette {
+    %option no_symbol_prefixing
 
     uword vera_palette_ptr
-    ubyte c
+    ubyte cc
 
     sub set_color(ubyte index, uword color) {
         vera_palette_ptr = $fa00+(index as uword * 2)
@@ -79,13 +80,13 @@ palette {
     sub set_grayscale() {
         vera_palette_ptr = $fa00
         repeat 16 {
-            c=0
+            cc=0
             repeat 16 {
-                cx16.vpoke(1, vera_palette_ptr, c)
+                cx16.vpoke(1, vera_palette_ptr, cc)
                 vera_palette_ptr++
-                cx16.vpoke(1, vera_palette_ptr, c)
+                cx16.vpoke(1, vera_palette_ptr, cc)
                 vera_palette_ptr++
-                c += $11
+                cc += $11
             }
         }
     }
@@ -150,11 +151,11 @@ palette {
     sub set_c64pepto() {
         vera_palette_ptr = $fa00
         repeat 16 {
-            for c in 0 to 15 {
-                uword cc = C64_colorpalette_pepto[c]
-                cx16.vpoke(1, vera_palette_ptr, lsb(cc))     ; G, B
+            for cc in 0 to 15 {
+                uword ccp = C64_colorpalette_pepto[cc]
+                cx16.vpoke(1, vera_palette_ptr, lsb(ccp))     ; G, B
                 vera_palette_ptr++
-                cx16.vpoke(1, vera_palette_ptr, msb(cc))     ; R
+                cx16.vpoke(1, vera_palette_ptr, msb(ccp))     ; R
                 vera_palette_ptr++
             }
         }
@@ -163,11 +164,11 @@ palette {
     sub set_c64light() {
         vera_palette_ptr = $fa00
         repeat 16 {
-            for c in 0 to 15 {
-                uword cc = C64_colorpalette_light[c]
-                cx16.vpoke(1, vera_palette_ptr, lsb(cc))     ; G, B
+            for cc in 0 to 15 {
+                uword ccp = C64_colorpalette_light[cc]
+                cx16.vpoke(1, vera_palette_ptr, lsb(ccp))     ; G, B
                 vera_palette_ptr++
-                cx16.vpoke(1, vera_palette_ptr, msb(cc))     ; R
+                cx16.vpoke(1, vera_palette_ptr, msb(ccp))     ; R
                 vera_palette_ptr++
             }
         }
@@ -176,11 +177,11 @@ palette {
     sub set_c64dark() {
         vera_palette_ptr = $fa00
         repeat 16 {
-            for c in 0 to 15 {
-                uword cc = C64_colorpalette_dark[c]
-                cx16.vpoke(1, vera_palette_ptr, lsb(cc))     ; G, B
+            for cc in 0 to 15 {
+                uword ccp = C64_colorpalette_dark[cc]
+                cx16.vpoke(1, vera_palette_ptr, lsb(ccp))     ; G, B
                 vera_palette_ptr++
-                cx16.vpoke(1, vera_palette_ptr, msb(cc))     ; R
+                cx16.vpoke(1, vera_palette_ptr, msb(ccp))     ; R
                 vera_palette_ptr++
             }
         }

compiler/res/prog8lib/cx16/psg.p8

@@ -1,6 +1,8 @@
 %import syslib
 
 psg {
+    %option no_symbol_prefixing
+
     ; $1F9C0 - $1F9FF 	16 blocks of 4 PSG registers (16 voices)
     ; 00  frequency word LSB
     ; 01  frequency word MSB.    freqword = HERZ / 0.3725290298461914
@@ -14,11 +16,11 @@ psg {
     const ubyte LEFT     = %01000000
     const ubyte RIGHT    = %10000000
 
-    sub voice(ubyte voice_num, ubyte channel, ubyte volume, ubyte waveform, ubyte pulsewidth) {
+    sub voice(ubyte voice_num, ubyte channel, ubyte vol, ubyte waveform, ubyte pulsewidth) {
         ; -- Enables a 'voice' on the PSG.
         ;    voice_num = 0-15, the voice number.
         ;    channel = either LEFT or RIGHT or (LEFT|RIGHT). Specifies the stereo channel(s) to use.
-        ;    volume = 0-63, the starting volume for the voice
+        ;    vol = 0-63, the starting volume for the voice
         ;    waveform = one of PULSE,SAWTOOTH,TRIANGLE,NOISE.
         ;    pulsewidth = 0-63.  Specifies the pulse width for waveform=PULSE.
         envelope_states[voice_num] = 255
@@ -28,11 +30,11 @@ psg {
         cx16.VERA_ADDR_L = lsb(cx16.r0)
         cx16.VERA_ADDR_M = msb(cx16.r0)
         cx16.VERA_ADDR_H = 1
-        cx16.VERA_DATA0 = channel | volume
+        cx16.VERA_DATA0 = channel | vol
         cx16.VERA_ADDR_L++
         cx16.VERA_DATA0 = waveform | pulsewidth
-        envelope_volumes[voice_num] = mkword(volume, 0)
-        envelope_maxvolumes[voice_num] = volume
+        envelope_volumes[voice_num] = mkword(vol, 0)
+        envelope_maxvolumes[voice_num] = vol
         sys.irqsafe_clear_irqd()
     }
 

compiler/res/prog8lib/cx16/sprites.p8

@@ -0,0 +1,110 @@
+; Simple routines to control sprites.
+; They're not written for high performance, but for simplicity.
+; That's why they control 1 sprite at a time. The exception is pos_batch().
+; which is quite efficient to update sprite positions of multiple sprites in one go.
+
+; note: sprites z-order will be in front of all layers.
+; note: collision mask is not supported here yet.
+
+sprites {
+    const ubyte SIZE_8  = 0
+    const ubyte SIZE_16 = 1
+    const ubyte SIZE_32 = 2
+    const ubyte SIZE_64 = 3
+    const ubyte COLORS_16 = 0
+    const ubyte COLORS_256 = 128
+    const uword VERA_SPRITEREGS = $fc00     ; $1fc00
+    uword @zp sprite_reg
+
+    sub init(ubyte spritenum,
+             ubyte databank, uword dataaddr,
+             ubyte width_flag, ubyte height_flag,
+             ubyte colors_flag, ubyte palette_offset_idx) {
+        hide(spritenum)
+        cx16.VERA_DC_VIDEO |= %01000000             ; enable sprites globally
+        dataaddr >>= 5
+        dataaddr |= (databank as uword)<<11
+        sprite_reg = VERA_SPRITEREGS + spritenum*$0008
+        cx16.vpoke(1, sprite_reg, lsb(dataaddr))                    ; address 12:5
+        cx16.vpoke(1, sprite_reg+1, colors_flag | msb(dataaddr))    ; 4 bpp + address 16:13
+        cx16.vpoke(1, sprite_reg+6, %00001100)                      ; z depth %11 = in front of both layers, no flips
+        cx16.vpoke(1, sprite_reg+7, height_flag<<6 | width_flag<<4 | palette_offset_idx>>4) ; 64x64 pixels, palette offset
+    }
+
+    sub data(ubyte spritenum, ubyte bank, uword addr) {
+        addr >>= 5
+        addr |= (bank as uword)<<11
+        sprite_reg = VERA_SPRITEREGS + spritenum*$0008
+        cx16.vpoke(1, sprite_reg, lsb(addr))                    ; address 12:5
+        cx16.vpoke_mask(1, sprite_reg+1, %11110000, msb(addr))    ; address 16:13
+    }
+
+    sub pos(ubyte spritenum, word xpos, word ypos) {
+        sprite_reg = VERA_SPRITEREGS + 2 + spritenum*$0008
+        cx16.vpoke(1, sprite_reg, lsb(xpos))
+        cx16.vpoke(1, sprite_reg+1, msb(xpos))
+        cx16.vpoke(1, sprite_reg+2, lsb(ypos))
+        cx16.vpoke(1, sprite_reg+3, msb(ypos))
+    }
+
+    sub pos_batch(ubyte first_spritenum, ubyte num_sprites, uword xpositions_ptr, uword ypositions_ptr) {
+        ; -- note: the x and y positions word arrays must be regular arrays, they cannot be @split arrays!
+        sprite_reg = VERA_SPRITEREGS + 2 + first_spritenum*$0008
+        cx16.vaddr_autoincr(1, sprite_reg, 0, 8)
+        cx16.vaddr_autoincr(1, sprite_reg+1, 1, 8)
+        repeat num_sprites {
+            cx16.VERA_DATA0 = @(xpositions_ptr)
+            xpositions_ptr ++
+            cx16.VERA_DATA1 = @(xpositions_ptr)
+            xpositions_ptr ++
+        }
+        sprite_reg += 2
+        cx16.vaddr_autoincr(1, sprite_reg, 0, 8)
+        cx16.vaddr_autoincr(1, sprite_reg+1, 1, 8)
+        repeat num_sprites {
+            cx16.VERA_DATA0 = @(ypositions_ptr)
+            ypositions_ptr ++
+            cx16.VERA_DATA1 = @(ypositions_ptr)
+            ypositions_ptr ++
+        }
+    }
+
+    sub setx(ubyte spritenum, word xpos) {
+        sprite_reg = VERA_SPRITEREGS + 2 + spritenum*$0008
+        cx16.vpoke(1, sprite_reg, lsb(xpos))
+        cx16.vpoke(1, sprite_reg+1, msb(xpos))
+    }
+
+    sub sety(ubyte spritenum, word ypos) {
+        sprite_reg = VERA_SPRITEREGS + 4 + spritenum*$0008
+        cx16.vpoke(1, sprite_reg, lsb(ypos))
+        cx16.vpoke(1, sprite_reg+1, msb(ypos))
+    }
+
+    sub getx(ubyte spritenum) -> word {
+        sprite_reg = VERA_SPRITEREGS + 2 + spritenum*$0008
+        return mkword(cx16.vpeek(1, sprite_reg+1), cx16.vpeek(1, sprite_reg)) as word
+    }
+
+    sub gety(ubyte spritenum) -> word {
+        sprite_reg = VERA_SPRITEREGS + 4 + spritenum*$0008
+        return mkword(cx16.vpeek(1, sprite_reg+1), cx16.vpeek(1, sprite_reg)) as word
+    }
+
+    sub hide(ubyte spritenum) {
+        pos(spritenum, -64, -64)
+    }
+
+    sub flipx(ubyte spritenum, bool flipped) {
+        cx16.vpoke_mask(1, VERA_SPRITEREGS + 6 + spritenum*$0008, %11111110, flipped)
+    }
+
+    sub flipy(ubyte spritenum, bool flipped) {
+        cx16.vpoke_mask(1, VERA_SPRITEREGS + 6 + spritenum*$0008, %11111101, flipped<<1)
+    }
+
+    sub set_palette_offset(ubyte spritenum, ubyte offset) {
+        sprite_reg = VERA_SPRITEREGS + 7 + spritenum*$0008
+        cx16.vpoke_mask(1, sprite_reg, %11110000, offset>>4)
+    }
+}

compiler/res/prog8lib/cx16/syslib.p8

@@ -5,6 +5,8 @@ cbm {
 
     ; Commodore (CBM) common variables, vectors and kernal routines
 
+    %option no_symbol_prefixing
+
 ; STROUT --> use txt.print
 ; CLEARSCR -> use txt.clear_screen
 ; HOMECRSR -> use txt.home or txt.plot
@@ -36,7 +38,7 @@ romsub $FFC6 = CHKIN(ubyte logical @ X) clobbers(A,X) -> bool @Pc    ; (via 798
 romsub $FFC9 = CHKOUT(ubyte logical @ X) clobbers(A,X)          ; (via 800 ($320)) define an output channel
 romsub $FFCC = CLRCHN() clobbers(A,X)                           ; (via 802 ($322)) restore default devices
 romsub $FFCF = CHRIN() clobbers(X, Y) -> ubyte @ A   ; (via 804 ($324)) input a character (for keyboard, read a whole line from the screen) A=byte read.
-romsub $FFD2 = CHROUT(ubyte char @ A)                           ; (via 806 ($326)) output a character
+romsub $FFD2 = CHROUT(ubyte character @ A)                           ; (via 806 ($326)) output a character
 romsub $FFD5 = LOAD(ubyte verify @ A, uword address @ XY) -> bool @Pc, ubyte @ A, uword @ XY     ; (via 816 ($330)) load from device
 romsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) clobbers (X, Y) -> bool @ Pc, ubyte @ A       ; (via 818 ($332)) save to a device
 romsub $FFDB = SETTIM(ubyte low @ A, ubyte middle @ X, ubyte high @ Y)      ; set the software clock
@@ -51,25 +53,21 @@ romsub $FFF3 = IOBASE() -> uword @ XY                           ; read base addr
 
 ; ---- utility
 
-asmsub STOP2() -> ubyte @A  {
+asmsub STOP2() clobbers(X) -> ubyte @A  {
     ; -- check if STOP key was pressed, returns true if so.  More convenient to use than STOP() because that only sets the carry status flag.
     %asm {{
-        phx
         jsr  cbm.STOP
         beq  +
-        plx
         lda  #0
         rts
-+       plx
-        lda  #1
++       lda  #1
         rts
     }}
 }
 
-asmsub RDTIM16() -> uword @AY {
+asmsub RDTIM16() clobbers(X) -> uword @AY {
     ; --  like RDTIM() but only returning the lower 16 bits in AY for convenience. Also avoids ram bank issue for irqs.
     %asm {{
-        phx
         php
         sei
         jsr  cbm.RDTIM
@@ -79,7 +77,6 @@ asmsub RDTIM16() -> uword @AY {
         txa
         tay
         pla
-        plx
         rts
     }}
 }
@@ -88,6 +85,8 @@ asmsub RDTIM16() -> uword @AY {
 
 cx16 {
 
+    %option no_symbol_prefixing
+
 ; irq, system and hardware vectors:
     &uword  IERROR      = $0300
     &uword  IMAIN       = $0302
@@ -324,7 +323,7 @@ cx16 {
 romsub $ff4a = close_all(ubyte device @A)  clobbers(A,X,Y)
 romsub $ff59 = lkupla(ubyte la @A)  clobbers(A,X,Y)
 romsub $ff5c = lkupsa(ubyte sa @Y)  clobbers(A,X,Y)
-romsub $ff5f = screen_mode(ubyte mode @A, bool getCurrent @Pc)  clobbers(A, X, Y) -> bool @Pc
+romsub $ff5f = screen_mode(ubyte mode @A, bool getCurrent @Pc)  clobbers(X, Y) -> ubyte @A, bool @Pc        ; note: X,Y size result is not supported, use SCREEN or get_screen_mode routine for that
 romsub $ff62 = screen_set_charset(ubyte charset @A, uword charsetptr @XY)  clobbers(A,X,Y)      ; incompatible with C128  dlchr()
 ; not yet supported: romsub $ff65 = pfkey()  clobbers(A,X,Y)
 romsub $ff6e = jsrfar()  ; following word = address to call, byte after that=rom/ram bank it is in
@@ -347,8 +346,8 @@ romsub $ff35 = GRAPH_draw_oval(uword x @R0, uword y @R1, uword width @R2, uword
 romsub $ff38 = GRAPH_draw_image(uword x @R0, uword y @R1, uword ptr @R2, uword width @R3, uword height @R4)  clobbers(A,X,Y)
 romsub $ff3b = GRAPH_set_font(uword fontptr @R0)  clobbers(A,X,Y)
 romsub $ff3e = GRAPH_get_char_size(ubyte baseline @A, ubyte width @X, ubyte height_or_style @Y, bool is_control @Pc)  clobbers(A,X,Y)
-romsub $ff41 = GRAPH_put_char(uword x @R0, uword y @R1, ubyte char @A)  clobbers(A,X,Y)
-romsub $ff41 = GRAPH_put_next_char(ubyte char @A)  clobbers(A,X,Y)     ; alias for the routine above that doesn't reset the position of the initial character
+romsub $ff41 = GRAPH_put_char(uword x @R0, uword y @R1, ubyte character @A)  clobbers(A,X,Y)
+romsub $ff41 = GRAPH_put_next_char(ubyte character @A)  clobbers(A,X,Y)     ; alias for the routine above that doesn't reset the position of the initial character
 
 ; framebuffer
 romsub $fef6 = FB_init()  clobbers(A,X,Y)
@@ -378,7 +377,7 @@ romsub $fee7 = memory_copy(uword source @R0, uword target @R1, uword num_bytes @
 romsub $feea = memory_crc(uword address @R0, uword num_bytes @R1)  clobbers(A,X,Y) -> uword @R2
 romsub $feed = memory_decompress(uword input @R0, uword output @R1)  clobbers(A,X,Y) -> uword @R1       ; last address +1 is result in R1
 romsub $fedb = console_init(uword x @R0, uword y @R1, uword width @R2, uword height @R3)  clobbers(A,X,Y)
-romsub $fede = console_put_char(ubyte char @A, bool wrapping @Pc)  clobbers(A,X,Y)
+romsub $fede = console_put_char(ubyte character @A, bool wrapping @Pc)  clobbers(A,X,Y)
 romsub $fee1 = console_get_char()  clobbers(X,Y) -> ubyte @A
 romsub $fed8 = console_put_image(uword pointer @R0, uword width @R1, uword height @R2)  clobbers(A,X,Y)
 romsub $fed5 = console_set_paging_message(uword msgptr @R0)  clobbers(A,X,Y)
@@ -396,6 +395,7 @@ romsub $febd = kbdbuf_peek() -> ubyte @A, ubyte @X     ; key in A, queue length
 romsub $febd = kbdbuf_peek2() -> uword @AX             ; alternative to above to not have the hassle to deal with multiple return values
 romsub $fec0 = kbdbuf_get_modifiers() -> ubyte @A
 romsub $fec3 = kbdbuf_put(ubyte key @A) clobbers(X)
+romsub $fed2 = keymap(uword identifier @XY, bool read @Pc) -> bool @Pc
 romsub $ff68 = mouse_config(ubyte shape @A, ubyte resX @X, ubyte resY @Y)  clobbers (A, X, Y)
 romsub $ff6b = mouse_get(ubyte zpdataptr @X) -> ubyte @A
 romsub $ff71 = mouse_scan()  clobbers(A, X, Y)
@@ -403,14 +403,33 @@ romsub $ff53 = joystick_scan()  clobbers(A, X, Y)
 romsub $ff56 = joystick_get(ubyte joynr @A) -> ubyte @A, ubyte @X, ubyte @Y
 romsub $ff56 = joystick_get2(ubyte joynr @A) clobbers(Y) -> uword @AX   ; alternative to above to not have the hassle to deal with multiple return values
 
-; Audio (bank 10)
-romsub $C04B = psg_init() clobbers(A,X,Y)
+; Audio (rom bank 10)
+romsub $C04B = psg_init() clobbers(A,X,Y)                              ; (re)init Vera PSG
 romsub $C063 = ym_init() clobbers(A,X,Y) -> bool @Pc                   ; (re)init YM chip
 romsub $C066 = ym_loaddefpatches() clobbers(A,X,Y) -> bool @Pc         ; load default YM patches
-romsub $C09F = audio_init() clobbers(A,X,Y) -> bool @Pc                ; (re)initialize PSG and YM audio chips
+romsub $C09F = audio_init() clobbers(A,X,Y) -> bool @Pc                ; (re)initialize both vera PSG and YM audio chips
 ; TODO: add more of the audio routines?
 
 
+asmsub set_screen_mode(ubyte mode @A) clobbers(A,X,Y) -> bool @Pc {
+    ; -- convenience wrapper for screen_mode() to just set a new mode (and return success)
+    %asm {{
+        clc
+        jmp  screen_mode
+    }}
+}
+
+asmsub get_screen_mode() -> byte @A, byte @X, byte @Y {
+    ; -- convenience wrapper for screen_mode() to just get the current mode in A, and size in characters in X+Y
+    ;    this does need a piece of inlined asm to call it ans store the result values if you call this from prog8 code
+    ;    Note: you can also just do the SEC yourself and simply call screen_mode() directly,
+    ;          or use the existing SCREEN kernal routine for just getting the size in characters.
+    %asm {{
+        sec
+        jmp  screen_mode
+    }}
+}
+
 asmsub kbdbuf_clear() {
     ; -- convenience helper routine to clear the keyboard buffer
     %asm {{
@@ -431,15 +450,12 @@ asmsub mouse_config2(ubyte shape @A) clobbers (A, X, Y) {
     }}
 }
 
-asmsub mouse_pos() -> ubyte @A {
+asmsub mouse_pos() clobbers(X) -> ubyte @A {
     ; -- short wrapper around mouse_get() kernal routine:
     ; -- gets the position of the mouse cursor in cx16.r0 and cx16.r1 (x/y coordinate), returns mouse button status.
     %asm {{
-        phx
         ldx  #cx16.r0
-        jsr  cx16.mouse_get
-        plx
-        rts
+        jmp  cx16.mouse_get
     }}
 }
 
@@ -477,7 +493,7 @@ inline asmsub getrambank() -> ubyte @A {
     }}
 }
 
-asmsub numbanks() -> uword @AY {
+asmsub numbanks() clobbers(X) -> uword @AY {
     ; -- Returns the number of available RAM banks according to the kernal (each bank is 8 Kb).
     ;    Note that the number of such banks can be 256 so a word is returned.
     ;    But just looking at the A register (the LSB of the result word) could suffice if you know that A=0 means 256 banks:
@@ -485,15 +501,13 @@ asmsub numbanks() -> uword @AY {
     ;    Kernal's MEMTOP routine reports 0 in this case but that doesn't mean 'zero banks', instead it means 256 banks,
     ;    as there is no X16 without at least 1 page of banked RAM. So this routine returns 256 instead of 0.
     %asm {{
-        phx
         sec
         jsr  cbm.MEMTOP
         ldy  #0
         cmp  #0
         bne  +
         iny
-+       plx
-        rts
++       rts
     }}
 }
 
@@ -502,7 +516,8 @@ asmsub vpeek(ubyte bank @A, uword address @XY) -> ubyte @A {
         ;    note: inefficient when reading multiple sequential bytes!
         %asm {{
                 pha
-                lda  #1
+                lda  cx16.VERA_CTRL
+                ora  #1
                 sta  cx16.VERA_CTRL
                 pla
                 and  #1
@@ -516,6 +531,8 @@ asmsub vpeek(ubyte bank @A, uword address @XY) -> ubyte @A {
 
 asmsub vaddr(ubyte bank @A, uword address @R0, ubyte addrsel @R1, byte autoIncrOrDecrByOne @Y) clobbers(A) {
         ; -- setup the VERA's data address register 0 or 1
+        ;    with optional auto increment or decrement of 1.
+        ;    Note that the vaddr_autoincr() and vaddr_autodecr() routines allow to set all possible strides, not just 1.
         %asm {{
             and  #1
             pha
@@ -539,6 +556,105 @@ asmsub vaddr(ubyte bank @A, uword address @R0, ubyte addrsel @R1, byte autoIncrO
         }}
 }
 
+asmsub vaddr_clone(ubyte port @A) clobbers (A,X,Y) {
+    ; -- clones Vera addresses from the given source port to the other one.
+    ;    leaves CTRL on the destination port.
+    %asm {{
+        sta  VERA_CTRL
+        ldx  VERA_ADDR_L
+        ldy  VERA_ADDR_H
+        phy
+        ldy  VERA_ADDR_M
+        eor  #1
+        sta  VERA_CTRL
+        stx  VERA_ADDR_L
+        sty  VERA_ADDR_M
+        ply
+        sty  VERA_ADDR_H
+        rts
+    }}
+}
+
+asmsub vaddr_autoincr(ubyte bank @A, uword address @R0, ubyte addrsel @R1, uword autoIncrAmount @R2) clobbers(A,Y) {
+        ; -- setup the VERA's data address register 0 or 1
+        ;    including setting up optional auto increment amount.
+        ;    Specifiying an unsupported amount results in amount of zero. See the Vera docs about what amounts are possible.
+        %asm {{
+            jsr  _setup
+            lda  cx16.r2H
+            ora  cx16.r2L
+            beq  +
+            jsr  _determine_incr_bits
++           ora  P8ZP_SCRATCH_REG
+            sta  cx16.VERA_ADDR_H
+            rts
+
+_setup      and  #1
+            sta  P8ZP_SCRATCH_REG
+            lda  cx16.r1
+            and  #1
+            sta  cx16.VERA_CTRL
+            lda  cx16.r0
+            sta  cx16.VERA_ADDR_L
+            lda  cx16.r0+1
+            sta  cx16.VERA_ADDR_M
+            rts
+
+_determine_incr_bits
+            lda  cx16.r2H
+            bne  _large
+            lda  cx16.r2L
+            ldy  #13
+-           cmp  _strides_lsb,y
+            beq  +
+            dey
+            bpl  -
++           tya
+            asl  a
+            asl  a
+            asl  a
+            asl  a
+            rts
+_large      ora  cx16.r2L
+            cmp  #1         ; 256
+            bne  +
+            lda  #9<<4
+            rts
++           cmp  #2         ; 512
+            bne  +
+            lda  #10<<4
+            rts
++           cmp  #65        ; 320
+            bne  +
+            lda  #14<<4
+            rts
++           cmp  #130       ; 640
+            bne  +
+            lda  #15<<4
+            rts
++           lda  #0
+            rts
+_strides_lsb    .byte   0,1,2,4,8,16,32,64,128,255,255,40,80,160,255,255
+        }}
+}
+
+asmsub vaddr_autodecr(ubyte bank @A, uword address @R0, ubyte addrsel @R1, uword autoDecrAmount @R2) clobbers(A,Y) {
+        ; -- setup the VERA's data address register 0 or 1
+        ;    including setting up optional auto decrement amount.
+        ;    Specifiying an unsupported amount results in amount of zero. See the Vera docs about what amounts are possible.
+        %asm {{
+            jsr  vaddr_autoincr._setup
+            lda  cx16.r2H
+            ora  cx16.r2L
+            beq  +
+            jsr  vaddr_autoincr._determine_incr_bits
+            ora  #%00001000         ; autodecrement
++           ora  P8ZP_SCRATCH_REG
+            sta  cx16.VERA_ADDR_H
+            rts
+        }}
+}
+
 asmsub vpoke(ubyte bank @A, uword address @R0, ubyte value @Y) clobbers(A) {
     ; -- write a single byte to VERA's video memory
     ;    note: inefficient when writing multiple sequential bytes!
@@ -629,6 +745,33 @@ asmsub vpoke_mask(ubyte bank @A, uword address @R0, ubyte mask @X, ubyte value @
     }}
 }
 
+asmsub save_virtual_registers() clobbers(A,Y) {
+    %asm {{
+        ldy  #31
+-       lda  cx16.r0,y
+        sta  _cx16_vreg_storage,y
+        dey
+        bpl  -
+        rts
+
+_cx16_vreg_storage
+        .word 0,0,0,0,0,0,0,0
+        .word 0,0,0,0,0,0,0,0
+    }}
+}
+
+asmsub restore_virtual_registers() clobbers(A,Y) {
+    %asm {{
+        ldy  #31
+-       lda  save_virtual_registers._cx16_vreg_storage,y
+        sta  cx16.r0,y
+        dey
+        bpl  -
+        rts
+    }}
+}
+
+
 asmsub save_vera_context() clobbers(A) {
     ; -- use this at the start of your IRQ handler if it uses Vera registers, to save the state
     %asm {{
@@ -683,6 +826,8 @@ asmsub restore_vera_context() clobbers(A) {
 sys {
     ; ------- lowlevel system routines --------
 
+    %option no_symbol_prefixing
+
     const ubyte target = 16         ;  compilation target specifier.  64 = C64,  128 = C128,  16 = CommanderX16.
 
 asmsub  init_system()  {
@@ -712,12 +857,13 @@ asmsub  init_system()  {
         lda  #$90       ; black
         jsr  cbm.CHROUT
         lda  #1
-        sta  $00        ; select ram bank 1
         jsr  cbm.CHROUT ; swap fg/bg
         lda  #$9e       ; yellow
         jsr  cbm.CHROUT
         lda  #147       ; clear screen
         jsr  cbm.CHROUT
+        lda  #PROG8_VARSHIGH_RAMBANK
+        sta  $00    ; select ram bank
         lda  #0
         tax
         tay
@@ -735,6 +881,8 @@ asmsub  init_system_phase2()  {
         sta  restore_irq._orig_irqvec
         lda  cx16.CINV+1
         sta  restore_irq._orig_irqvec+1
+        lda  #PROG8_VARSHIGH_RAMBANK
+        sta  $00    ; select ram bank
         cli
         rts
     }}
@@ -800,10 +948,6 @@ _irq_handler_init
 		sta  IRQ_SCRATCH_ZPWORD2
 		lda  P8ZP_SCRATCH_W2+1
 		sta  IRQ_SCRATCH_ZPWORD2+1
-		; Set X to the bottom 32 bytes of the evaluation stack, to HOPEFULLY not clobber it.
-		; This leaves 128-32=96 stack entries for the main program, and 32 stack entries for the IRQ handler.
-		; We assume IRQ handlers don't contain complex expressions taking up more than that.
-		ldx  #32
 		cld
 		rts
 
@@ -906,8 +1050,8 @@ asmsub  set_rasterline(uword line @AY) {
         %asm {{
             sei
             ldx #$42
-            ldy #1
-            tya
+            ldy #2
+            lda #0
             jsr  cx16.i2c_write_byte
             bra  *
         }}
@@ -923,19 +1067,17 @@ asmsub  set_rasterline(uword line @AY) {
         void cx16.i2c_write_byte($42, $05, activity)
     }
 
-    asmsub wait(uword jiffies @AY) {
+    asmsub wait(uword jiffies @AY) clobbers(X) {
         ; --- wait approximately the given number of jiffies (1/60th seconds) (N or N+1)
         ;     note: the system irq handler has to be active for this to work as it depends on the system jiffy clock
         ;     note: this routine cannot be used from inside a irq handler
         %asm {{
-            phx
             sta  P8ZP_SCRATCH_W1
             sty  P8ZP_SCRATCH_W1+1
 
 _loop       lda  P8ZP_SCRATCH_W1
             ora  P8ZP_SCRATCH_W1+1
             bne  +
-            plx
             rts
 
 +           sei

compiler/res/prog8lib/cx16/textio.p8

@@ -6,6 +6,8 @@
 
 txt {
 
+    %option no_symbol_prefixing
+
 const ubyte DEFAULT_WIDTH = 80
 const ubyte DEFAULT_HEIGHT = 60
 
@@ -40,11 +42,10 @@ asmsub column(ubyte col @A) clobbers(A, X, Y) {
     }}
 }
 
-asmsub  fill_screen (ubyte char @ A, ubyte color @ Y) clobbers(A)  {
+asmsub  fill_screen (ubyte character @ A, ubyte color @ Y) clobbers(A, X)  {
 	; ---- fill the character screen with the given fill character and character color.
 	%asm {{
         sty  _ly+1
-        phx
         pha
         jsr  cbm.SCREEN             ; get dimensions in X/Y
         txa
@@ -73,8 +74,7 @@ _ly     ldy  #1                     ; modified
         stz  cx16.VERA_ADDR_L
         inc  cx16.VERA_ADDR_M       ; next line
         bra  _lx
-+       plx
-        rts
++       rts
 
 set_vera_textmatrix_addresses:
         stz  cx16.VERA_CTRL
@@ -88,11 +88,10 @@ set_vera_textmatrix_addresses:
         }}
 }
 
-asmsub  clear_screenchars (ubyte char @ A) clobbers(Y)  {
+asmsub  clear_screenchars (ubyte character @ A) clobbers(X, Y)  {
 	; ---- clear the character screen with the given fill character (leaves colors)
 	;      (assumes screen matrix is at the default address)
 	%asm {{
-        phx
         pha
         jsr  cbm.SCREEN             ; get dimensions in X/Y
         txa
@@ -114,16 +113,14 @@ _lx     ldx  #0                     ; modified
         stz  cx16.VERA_ADDR_L
         inc  cx16.VERA_ADDR_M       ; next line
         bra  _lx
-+       plx
-        rts
++       rts
         }}
 }
 
-asmsub  clear_screencolors (ubyte color @ A) clobbers(Y)  {
+asmsub  clear_screencolors (ubyte color @ A) clobbers(X, Y)  {
 	; ---- clear the character screen colors with the given color (leaves characters).
 	;      (assumes color matrix is at the default address)
 	%asm {{
-        phx
         sta  _la+1
         jsr  cbm.SCREEN             ; get dimensions in X/Y
         txa
@@ -148,8 +145,7 @@ _la     lda  #0                     ; modified
         sta  cx16.VERA_ADDR_L
         inc  cx16.VERA_ADDR_M       ; next line
         bra  _lx
-+       plx
-        rts
++       rts
         }}
 }
 
@@ -190,11 +186,10 @@ sub iso_off() {
 }
 
 
-asmsub  scroll_left() clobbers(A, Y)  {
+asmsub  scroll_left() clobbers(A, X, Y)  {
 	; ---- scroll the whole screen 1 character to the left
 	;      contents of the rightmost column are unchanged, you should clear/refill this yourself
 	%asm {{
-	    phx
 	    jsr  cbm.SCREEN
 	    dex
 	    stx  _lx+1
@@ -231,16 +226,14 @@ _lx     ldx  #0                ; modified
 
         lda  #0
         sta  cx16.VERA_CTRL
-	    plx
 	    rts
 	}}
 }
 
-asmsub  scroll_right() clobbers(A)  {
+asmsub  scroll_right() clobbers(A,X,Y)  {
 	; ---- scroll the whole screen 1 character to the right
 	;      contents of the leftmost column are unchanged, you should clear/refill this yourself
 	%asm {{
-	    phx
 	    jsr  cbm.SCREEN
 	    dex
 	    stx  _lx+1
@@ -285,16 +278,14 @@ _lx     ldx  #0                 ; modified
 
         lda  #0
         sta  cx16.VERA_CTRL
-	    plx
 	    rts
 	}}
 }
 
-asmsub  scroll_up() clobbers(A, Y)  {
+asmsub  scroll_up() clobbers(A, X, Y)  {
 	; ---- scroll the whole screen 1 character up
 	;      contents of the bottom row are unchanged, you should refill/clear this yourself
 	%asm {{
-	    phx
 	    jsr  cbm.SCREEN
 	    stx  _nextline+1
 	    dey
@@ -335,16 +326,14 @@ _nextline
 
 +       lda  #0
         sta  cx16.VERA_CTRL
-	    plx
 	    rts
 	}}
 }
 
-asmsub  scroll_down() clobbers(A, Y)  {
+asmsub  scroll_down() clobbers(A, X, Y)  {
 	; ---- scroll the whole screen 1 character down
 	;      contents of the top row are unchanged, you should refill/clear this yourself
 	%asm {{
-	    phx
 	    jsr  cbm.SCREEN
 	    stx  _nextline+1
 	    dey
@@ -391,12 +380,11 @@ _nextline
 
 +       lda  #0
         sta  cx16.VERA_CTRL
-	    plx
 	    rts
 	}}
 }
 
-romsub $FFD2 = chrout(ubyte char @ A)    ; for consistency. You can also use cbm.CHROUT directly ofcourse.
+romsub $FFD2 = chrout(ubyte character @ A)    ; for consistency. You can also use cbm.CHROUT directly ofcourse.
 
 asmsub  print (str text @ AY) clobbers(A,Y)  {
 	; ---- print null terminated string from A/Y
@@ -416,10 +404,9 @@ asmsub  print (str text @ AY) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  print_ub0  (ubyte value @ A) clobbers(A,Y)  {
+asmsub  print_ub0  (ubyte value @ A) clobbers(A,X,Y)  {
 	; ---- print the ubyte in A in decimal form, with left padding 0s (3 positions total)
 	%asm {{
-		phx
 		jsr  conv.ubyte2decimal
 		pha
 		tya
@@ -427,16 +414,13 @@ asmsub  print_ub0  (ubyte value @ A) clobbers(A,Y)  {
 		pla
 		jsr  cbm.CHROUT
 		txa
-		jsr  cbm.CHROUT
-		plx
-		rts
+		jmp  cbm.CHROUT
 	}}
 }
 
-asmsub  print_ub  (ubyte value @ A) clobbers(A,Y)  {
+asmsub  print_ub  (ubyte value @ A) clobbers(A,X,Y)  {
 	; ---- print the ubyte in A in decimal form, without left padding 0s
 	%asm {{
-		phx
 		jsr  conv.ubyte2decimal
 _print_byte_digits
 		pha
@@ -452,16 +436,13 @@ _print_byte_digits
         beq  _ones
         jsr  cbm.CHROUT
 _ones   txa
-		jsr  cbm.CHROUT
-		plx
-		rts
+		jmp  cbm.CHROUT
 	}}
 }
 
-asmsub  print_b  (byte value @ A) clobbers(A,Y)  {
+asmsub  print_b  (byte value @ A) clobbers(A,X,Y)  {
 	; ---- print the byte in A in decimal form, without left padding 0s
 	%asm {{
-		phx
 		pha
 		cmp  #0
 		bpl  +
@@ -473,10 +454,9 @@ asmsub  print_b  (byte value @ A) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  print_ubhex  (ubyte value @ A, bool prefix @ Pc) clobbers(A,Y)  {
+asmsub  print_ubhex  (ubyte value @ A, bool prefix @ Pc) clobbers(A,X,Y)  {
 	; ---- print the ubyte in A in hex form (if Carry is set, a radix prefix '$' is printed as well)
 	%asm {{
-		phx
 		bcc  +
 		pha
 		lda  #'$'
@@ -485,16 +465,13 @@ asmsub  print_ubhex  (ubyte value @ A, bool prefix @ Pc) clobbers(A,Y)  {
 +		jsr  conv.ubyte2hex
 		jsr  cbm.CHROUT
 		tya
-		jsr  cbm.CHROUT
-		plx
-		rts
+		jmp  cbm.CHROUT
 	}}
 }
 
-asmsub  print_ubbin  (ubyte value @ A, bool prefix @ Pc) clobbers(A,Y)  {
+asmsub  print_ubbin  (ubyte value @ A, bool prefix @ Pc) clobbers(A,X,Y)  {
 	; ---- print the ubyte in A in binary form (if Carry is set, a radix prefix '%' is printed as well)
 	%asm {{
-		phx
 		sta  P8ZP_SCRATCH_B1
 		bcc  +
 		lda  #'%'
@@ -507,12 +484,11 @@ asmsub  print_ubbin  (ubyte value @ A, bool prefix @ Pc) clobbers(A,Y)  {
 +		jsr  cbm.CHROUT
 		dey
 		bne  -
-		plx
 		rts
 	}}
 }
 
-asmsub  print_uwbin  (uword value @ AY, bool prefix @ Pc) clobbers(A,Y)  {
+asmsub  print_uwbin  (uword value @ AY, bool prefix @ Pc) clobbers(A,X,Y)  {
 	; ---- print the uword in A/Y in binary form (if Carry is set, a radix prefix '%' is printed as well)
 	%asm {{
 		pha
@@ -524,7 +500,7 @@ asmsub  print_uwbin  (uword value @ AY, bool prefix @ Pc) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  print_uwhex  (uword value @ AY, bool prefix @ Pc) clobbers(A,Y)  {
+asmsub  print_uwhex  (uword value @ AY, bool prefix @ Pc) clobbers(A,X,Y)  {
 	; ---- print the uword in A/Y in hexadecimal form (4 digits)
 	;      (if Carry is set, a radix prefix '$' is printed as well)
 	%asm {{
@@ -537,10 +513,9 @@ asmsub  print_uwhex  (uword value @ AY, bool prefix @ Pc) clobbers(A,Y)  {
 	}}
 }
 
-asmsub  print_uw0  (uword value @ AY) clobbers(A,Y)  {
+asmsub  print_uw0  (uword value @ AY) clobbers(A,X,Y)  {
 	; ---- print the uword in A/Y in decimal form, with left padding 0s (5 positions total)
 	%asm {{
-	    phx
 		jsr  conv.uword2decimal
 		ldy  #0
 -		lda  conv.uword2decimal.decTenThousands,y
@@ -548,17 +523,14 @@ asmsub  print_uw0  (uword value @ AY) clobbers(A,Y)  {
 		jsr  cbm.CHROUT
 		iny
 		bne  -
-+		plx
-		rts
++		rts
 	}}
 }
 
-asmsub  print_uw  (uword value @ AY) clobbers(A,Y)  {
+asmsub  print_uw  (uword value @ AY) clobbers(A,X,Y)  {
 	; ---- print the uword in A/Y in decimal form, without left padding 0s
 	%asm {{
-	    phx
 		jsr  conv.uword2decimal
-		plx
 		ldy  #0
 -		lda  conv.uword2decimal.decTenThousands,y
 		beq  _allzero
@@ -579,7 +551,7 @@ _allzero
 	}}
 }
 
-asmsub  print_w  (word value @ AY) clobbers(A,Y)  {
+asmsub  print_w  (word value @ AY) clobbers(A,X,Y)  {
 	; ---- print the (signed) word in A/Y in decimal form, without left padding 0's
 	%asm {{
 		cpy  #0
@@ -702,13 +674,12 @@ asmsub  getclr  (ubyte col @A, ubyte row @Y) -> ubyte @ A {
 	}}
 }
 
-sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
+sub  setcc  (ubyte col, ubyte row, ubyte character, ubyte charcolor) {
 	; ---- set char+color at the given position on the screen
 	;      note: color handling is the same as on the C64: it only sets the foreground color and leaves the background color as is.
 	;            Use setcc2 if you want Cx-16 specific feature of setting both Bg+Fg colors (is faster as well).
 	%asm {{
-            phx
-            lda  column
+            lda  col
             asl  a
             tax
             ldy  row
@@ -720,7 +691,7 @@ sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
             ;clc
             adc  #>VERA_TEXTMATRIX_ADDR
             sta  cx16.VERA_ADDR_M
-            lda  char
+            lda  character
             sta  cx16.VERA_DATA0
             inc  cx16.VERA_ADDR_L
             lda  charcolor
@@ -730,18 +701,16 @@ sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
             and  #$f0
             ora  P8ZP_SCRATCH_B1
             sta  cx16.VERA_DATA0
-            plx
             rts
     }}
 }
 
-sub  setcc2  (ubyte column, ubyte row, ubyte char, ubyte colors)  {
+sub  setcc2  (ubyte col, ubyte row, ubyte character, ubyte colors)  {
 	; ---- set char+color at the given position on the screen
 	;      note: on the CommanderX16 this allows you to set both Fg and Bg colors;
 	;            use the high nybble in A to set the Bg color! Is a bit faster than setcc() too.
 	%asm {{
-            phx
-            lda  column
+            lda  col
             asl  a
             tax
             ldy  row
@@ -753,25 +722,19 @@ sub  setcc2  (ubyte column, ubyte row, ubyte char, ubyte colors)  {
             ; clc
             adc  #>VERA_TEXTMATRIX_ADDR
             sta  cx16.VERA_ADDR_M
-            lda  char
+            lda  character
             sta  cx16.VERA_DATA0
             inc  cx16.VERA_ADDR_L
             lda  colors
             sta  cx16.VERA_DATA0
-            plx
             rts
     }}
 }
 
-asmsub  plot  (ubyte col @ Y, ubyte row @ A) clobbers(A) {
-	; ---- safe wrapper around PLOT kernal routine, to save the X register.
+asmsub  plot  (ubyte col @ Y, ubyte row @ X) {
 	%asm  {{
-		phx
-		tax
 		clc
-		jsr  cbm.PLOT
-		plx
-		rts
+		jmp  cbm.PLOT
 	}}
 }
 

compiler/res/prog8lib/cx16logo.p8

@@ -3,6 +3,8 @@
 %import textio
 
 cx16logo {
+    %option no_symbol_prefixing
+
     sub logo_at(ubyte column, ubyte row) {
         uword strptr
         for strptr in logo_lines {

compiler/res/prog8lib/diskio.p8

@@ -5,6 +5,7 @@
 %import syslib
 
 diskio {
+    %option no_symbol_prefixing
 
     ubyte drivenumber = 8
 
@@ -41,12 +42,12 @@ diskio {
             ubyte high = cbm.CHRIN()
             txt.print_uw(mkword(high, low))
             txt.spc()
-            ubyte @zp char
+            ubyte @zp character
             repeat {
-                char = cbm.CHRIN()
-                if char==0
+                character = cbm.CHRIN()
+                if character==0
                     break
-                txt.chrout(char)
+                txt.chrout(character)
             }
             txt.nl()
             void cbm.CHRIN()     ; skip 2 bytes
@@ -84,8 +85,8 @@ io_error:
         if_cs
             goto io_error
 
-        repeat 6 {
-            void cbm.CHRIN()     ; skip the 6 prologue bytes
+        while cbm.CHRIN()!='"' {
+            ; skip up to entry name
         }
         if cbm.READST()!=0
             goto io_error
@@ -93,8 +94,14 @@ io_error:
         cx16.r0 = &list_filename
         repeat {
             @(cx16.r0) = cbm.CHRIN()
-            if @(cx16.r0)==0
+            if @(cx16.r0)=='"' {
+                @(cx16.r0) = ' '
+                while @(cx16.r0)==' ' and cx16.r0>=&diskio.list_filename {
+                    @(cx16.r0) = 0
+                    cx16.r0--
+                }
                 break
+            }
             cx16.r0++
         }
         okay = true
@@ -204,12 +211,12 @@ io_error:
 
             ; read the filename
             repeat {
-                ubyte char = cbm.CHRIN()
-                if char==0
+                ubyte character = cbm.CHRIN()
+                if character==0
                     break
-                if char=='\"'
+                if character=='\"'
                     break
-                @(nameptr) = char
+                @(nameptr) = character
                 nameptr++
             }
 
@@ -437,24 +444,22 @@ io_error:
         goto done
     }
 
-    sub save(uword filenameptr, uword address, uword size) -> bool {
+    sub save(uword filenameptr, uword start_address, uword savesize) -> bool {
         cbm.SETNAM(string.length(filenameptr), filenameptr)
         cbm.SETLFS(1, drivenumber, 0)
-        uword @shared end_address = address + size
+        uword @shared end_address = start_address + savesize
         cx16.r0L = 0
 
         %asm {{
-            lda  address
+            lda  start_address
             sta  P8ZP_SCRATCH_W1
-            lda  address+1
+            lda  start_address+1
             sta  P8ZP_SCRATCH_W1+1
-            stx  P8ZP_SCRATCH_REG
             lda  #<P8ZP_SCRATCH_W1
             ldx  end_address
             ldy  end_address+1
             jsr  cbm.SAVE
             php
-            ldx  P8ZP_SCRATCH_REG
             plp
         }}
 
@@ -481,7 +486,6 @@ io_error:
             secondary = 0
         cbm.SETLFS(1, drivenumber, secondary)
         %asm {{
-            stx  P8ZP_SCRATCH_REG
             lda  #0
             ldx  address_override
             ldy  address_override+1
@@ -489,7 +493,7 @@ io_error:
             bcs  +
             stx  cx16.r1
             sty  cx16.r1+1
-+           ldx  P8ZP_SCRATCH_REG
++
         }}
 
         cbm.CLRCHN()
@@ -500,16 +504,16 @@ io_error:
     ; Identical to load(), but DOES INCLUDE the first 2 bytes in the file.
     ; No program header is assumed in the file. Everything is loaded.
     ; See comments on load() for more details.
-    sub load_raw(uword filenameptr, uword address) -> uword {
+    sub load_raw(uword filenameptr, uword start_address) -> uword {
         ; read the 2 header bytes separately to skip them
         if not f_open(filenameptr)
             return 0
-        cx16.r1 = f_read(address, 2)
+        cx16.r1 = f_read(start_address, 2)
         f_close()
         if cx16.r1!=2
             return 0
-        address += 2
-        return load(filenameptr, address)
+        start_address += 2
+        return load(filenameptr, start_address)
     }
 
     sub delete(uword filenameptr) {

compiler/res/prog8lib/floats_functions.p8

@@ -1,11 +1,10 @@
 floats {
     ; the floating point functions shared across compiler targets
-    %option merge
+    %option merge, no_symbol_prefixing
 
 sub print_f(float value) {
 	; ---- prints the floating point value (without a newline).
 	%asm {{
-		stx  floats_store_reg
 		lda  #<value
 		ldy  #>value
 		jsr  MOVFM		; load float into fac1
@@ -18,8 +17,7 @@ sub print_f(float value) {
 		jsr  cbm.CHROUT
 		iny
 		bne  -
-+		ldx  floats_store_reg
-		rts
++		rts
 	}}
 }
 
@@ -44,10 +42,7 @@ sub sin(float angle) -> float {
         lda  #<angle
         ldy  #>angle
         jsr  MOVFM
-        stx  P8ZP_SCRATCH_REG
-        jsr  SIN
-        ldx  P8ZP_SCRATCH_REG
-        rts
+        jmp  SIN
     }}
 }
 
@@ -56,9 +51,7 @@ sub cos(float angle) -> float {
         lda  #<angle
         ldy  #>angle
         jsr  MOVFM
-        stx  P8ZP_SCRATCH_REG
-        jsr  COS
-        ldx  P8ZP_SCRATCH_REG
+        jmp  COS
         rts
     }}
 }
@@ -68,10 +61,7 @@ sub tan(float value) -> float {
         lda  #<value
         ldy  #>value
         jsr  MOVFM
-        stx  P8ZP_SCRATCH_REG
-        jsr  TAN
-        ldx  P8ZP_SCRATCH_REG
-        rts
+        jmp  TAN
     }}
 }
 
@@ -80,10 +70,7 @@ sub atan(float value) -> float {
         lda  #<value
         ldy  #>value
         jsr  MOVFM
-        stx  P8ZP_SCRATCH_REG
-        jsr  ATN
-        ldx  P8ZP_SCRATCH_REG
-        rts
+        jmp  ATN
     }}
 }
 
@@ -92,10 +79,7 @@ sub ln(float value) -> float {
         lda  #<value
         ldy  #>value
         jsr  MOVFM
-        stx  P8ZP_SCRATCH_REG
-        jsr  LOG
-        ldx  P8ZP_SCRATCH_REG
-        rts
+        jmp  LOG
     }}
 }
 
@@ -104,15 +88,12 @@ sub log2(float value) -> float {
         lda  #<value
         ldy  #>value
         jsr  MOVFM
-        stx  P8ZP_SCRATCH_REG
         jsr  LOG
         jsr  MOVEF
         lda  #<FL_LOG2_const
         ldy  #>FL_LOG2_const
         jsr  MOVFM
-        jsr  FDIVT
-        ldx  P8ZP_SCRATCH_REG
-        rts
+        jmp  FDIVT
     }}
 }
 
@@ -122,12 +103,9 @@ sub rad(float angle) -> float {
         lda  #<angle
         ldy  #>angle
         jsr  MOVFM
-        stx  P8ZP_SCRATCH_REG
         lda  #<_pi_div_180
         ldy  #>_pi_div_180
-        jsr  FMULT
-        ldx  P8ZP_SCRATCH_REG
-        rts
+        jmp  FMULT
 _pi_div_180	.byte 123, 14, 250, 53, 18		; pi / 180
     }}
 }
@@ -138,11 +116,9 @@ sub deg(float angle) -> float {
         lda  #<angle
         ldy  #>angle
         jsr  MOVFM
-        stx  P8ZP_SCRATCH_REG
         lda  #<_one_over_pi_div_180
         ldy  #>_one_over_pi_div_180
-        jsr  FMULT
-        ldx  P8ZP_SCRATCH_REG
+        jmp  FMULT
         rts
 _one_over_pi_div_180	.byte 134, 101, 46, 224, 211		; 1 / (pi * 180)
     }}
@@ -153,11 +129,8 @@ sub round(float value) -> float {
         lda  #<value
         ldy  #>value
         jsr  MOVFM
-        stx  P8ZP_SCRATCH_REG
         jsr  FADDH
-        jsr  INT
-        ldx  P8ZP_SCRATCH_REG
-        rts
+        jmp  INT
     }}
 }
 
@@ -166,10 +139,7 @@ sub floor(float value) -> float {
         lda  #<value
         ldy  #>value
         jsr  MOVFM
-        stx  P8ZP_SCRATCH_REG
-        jsr  INT
-        ldx  P8ZP_SCRATCH_REG
-        rts
+        jmp  INT
     }}
 }
 
@@ -179,7 +149,6 @@ sub ceil(float value) -> float {
         lda  #<value
         ldy  #>value
         jsr  MOVFM
-        stx  P8ZP_SCRATCH_REG
         ldx  #<fmath_float1
         ldy  #>fmath_float1
         jsr  MOVMF
@@ -192,8 +161,7 @@ sub ceil(float value) -> float {
         lda  #<FL_ONE_const
         ldy  #>FL_ONE_const
         jsr  FADD
-+       ldx  P8ZP_SCRATCH_REG
-        rts
++       rts
     }}
 }
 
@@ -202,14 +170,11 @@ sub rndseedf(float seed) {
         seed = -seed    ; make sure fp seed is always negative
 
     %asm {{
-        stx  floats_store_reg
         lda  #<seed
         ldy  #>seed
         jsr  MOVFM		; load float into fac1
         lda  #-1
-        jsr  floats.RND
-        ldx  floats_store_reg
-        rts
+        jmp  floats.RND
     }}
 }
 

compiler/res/prog8lib/math.p8

@@ -1,6 +1,8 @@
 ; Internal Math library routines - always included by the compiler
 
 math {
+    %option no_symbol_prefixing
+
 	%asminclude "library:math.asm"
 
     asmsub sin8u(ubyte angle @A) clobbers(Y) -> ubyte @A {
@@ -95,4 +97,337 @@ _sinecosR8	.char  trunc(127.0 * sin(range(180+45) * rad(360.0/180.0)))
             rts
         }}
     }
+
+    asmsub log2(ubyte value @A) -> ubyte @Y {
+        %asm {{
+            sta  P8ZP_SCRATCH_B1
+            lda  #$80
+            ldy  #7
+-           bit  P8ZP_SCRATCH_B1
+            beq  +
+            rts
++           dey
+            bne  +
+            rts
++           lsr  a
+            bne  -
+        }}
+    }
+
+    asmsub log2w(uword value @AY) -> ubyte @Y {
+        %asm {{
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            lda  #<$8000
+            sta  cx16.r0
+            lda  #>$8000
+            sta  cx16.r0+1
+            ldy  #15
+-           lda  P8ZP_SCRATCH_W1
+            and  cx16.r0
+            sta  P8ZP_SCRATCH_B1
+            lda  P8ZP_SCRATCH_W1+1
+            and  cx16.r0+1
+            ora  P8ZP_SCRATCH_B1
+            beq  +
+            rts
++           dey
+            bne  +
+            rts
++           lsr  cx16.r0+1
+            ror  cx16.r0
+            jmp  -
+        }}
+    }
+
+sub direction_sc(byte x1, byte y1, byte x2, byte y2) -> ubyte {
+    ; From a pair of signed coordinates around the origin, calculate discrete direction between 0 and 23 into A.
+    cx16.r0L = 3        ; quadrant
+    cx16.r1sL = x2-x1   ; xdelta
+    if_neg {
+        cx16.r0L--
+        cx16.r1sL = -cx16.r1sL
+    }
+    cx16.r2sL = y2-y1   ; ydelta
+    if_neg {
+        cx16.r0L-=2
+        cx16.r2sL = -cx16.r2sL
+    }
+    return direction_qd(cx16.r0L, cx16.r1L, cx16.r2L)
+}
+
+sub direction(ubyte x1, ubyte y1, ubyte x2, ubyte y2) -> ubyte {
+    ; From a pair of positive coordinates, calculate discrete direction between 0 and 23 into A.
+    cx16.r0L = 3        ; quadrant
+    if x2>=x1 {
+        cx16.r1L = x2-x1
+    } else {
+        cx16.r1L = x1-x2
+        cx16.r0L--
+    }
+    if y2>=y1 {
+        cx16.r2L = y2-y1
+    } else {
+        cx16.r2L = y1-y2
+        cx16.r0L -= 2
+    }
+    return direction_qd(cx16.r0L, cx16.r1L, cx16.r2L)
+}
+
+asmsub direction_qd(ubyte quadrant @A, ubyte xdelta @X, ubyte ydelta @Y) -> ubyte @A {
+    ;Arctan  https://github.com/dustmop/arctan24
+    ; From a pair of X/Y deltas (both >=0), and quadrant 0-3, calculate discrete direction between 0 and 23 into A.
+    ;  .reg:a @in  quadrant Number 0 to 3.
+    ;  .reg:x @in  x_delta Delta for x direction.
+    ;  .reg:y @in  y_delta Delta for y direction.
+    ; Returns A as the direction (0-23).
+
+    %asm {{
+x_delta = cx16.r0L
+y_delta = cx16.r1L
+quadrant = cx16.r2L
+half_value = cx16.r3L
+region_number = cx16.r4L
+small = cx16.r5L
+large = cx16.r5H
+
+  sta quadrant
+  sty y_delta
+  stx x_delta
+  cpx y_delta
+  bcs _XGreaterOrEqualY
+
+_XLessY:
+  lda #16
+  sta region_number
+  stx small
+  sty large
+  bne _DetermineRegion
+
+_XGreaterOrEqualY:
+  lda #0
+  sta region_number
+  stx large
+  sty small
+
+_DetermineRegion:
+  ; set A = small * 2.5
+  lda small
+  lsr a
+  sta half_value
+  lda small
+  asl a
+  bcs _SmallerQuotient
+  clc
+  adc half_value
+  bcs _SmallerQuotient
+  cmp large
+  bcc _LargerQuotient
+
+; S * 2.5 > L
+_SmallerQuotient:
+  ; set A = S * 1.25
+  lsr half_value
+  lda small
+  clc
+  adc half_value
+  cmp large
+  bcc _Region1 ; if S * 1.25 < L then goto Region1 (L / S > 1.25)
+  bcs _Region0 ;                                   (L / S < 1.25)
+
+; S * 2.5 < L
+_LargerQuotient:
+  ; set A = S * 7.5
+  lda small
+  asl a
+  asl a
+  asl a
+  bcs _Region2
+  sec
+  sbc half_value
+  cmp large
+  bcc _Region3 ; if S * 7.5 < L then goto Region3 (L / S > 7.5)
+  jmp _Region2 ;                                  (L / S < 7.5)
+
+_Region0:
+  ; L / S < 1.25. d=3,9,15,21
+  jmp _LookupResult
+
+_Region1:
+  ; 1.25 < L / S < 2.5. d=2,4,8,10,14,16,20,22
+  lda region_number
+  clc
+  adc #4
+  sta region_number
+  bpl _LookupResult
+
+_Region2:
+  ; 2.5 < L / S < 7.5. d=1,5,7,11,13,17,19,23
+  lda region_number
+  clc
+  adc #8
+  sta region_number
+  bpl _LookupResult
+
+_Region3:
+  ; 7.5 < L / S. d=0,6,12,18
+  lda region_number
+  clc
+  adc #12
+  sta region_number
+
+_LookupResult:
+  lda quadrant
+  clc
+  adc region_number
+  tax
+  lda _quadrant_region_to_direction,x
+  rts
+
+_quadrant_region_to_direction:
+  .byte  9, 3,15,21
+  .byte 10, 2,14,22
+  .byte 11, 1,13,23
+  .byte 12, 0,12, 0
+  .byte  9, 3,15,21
+  .byte  8, 4,16,20
+  .byte  7, 5,17,19
+  .byte  6, 6,18,18
+
+    }}
+}
+
+asmsub atan2(ubyte x1 @R0, ubyte y1 @R1, ubyte x2 @R2, ubyte y2 @R3) -> ubyte @A {
+    ;; Calculate the angle, in a 256-degree circle, between two points into A.
+    ;; The points (x1, y1) and (x2, y2) have to use *unsigned coordinates only* from the positive quadrant in the carthesian plane!
+    ;; https://www.codebase64.org/doku.php?id=base:8bit_atan2_8-bit_angle
+    ;; This uses 2 large lookup tables so uses a lot of memory but is super fast.
+
+    %asm {{
+
+x1 = cx16.r0L
+y1 = cx16.r1L
+x2 = cx16.r2L
+y2 = cx16.r3L
+octant = cx16.r4L			;; temporary zeropage variable
+
+		lda x1
+		sec
+		sbc x2
+		bcs *+4
+		eor #$ff
+		tax
+		rol octant
+
+		lda y1
+		sec
+		sbc y2
+		bcs *+4
+		eor #$ff
+		tay
+		rol octant
+
+		lda log2_tab,x
+		sec
+		sbc log2_tab,y
+		bcc *+4
+		eor #$ff
+		tax
+
+		lda octant
+		rol a
+		and #%111
+		tay
+
+		lda atan_tab,x
+		eor octant_adjust,y
+		rts
+
+octant_adjust
+		.byte %00111111		;; x+,y+,|x|>|y|
+		.byte %00000000		;; x+,y+,|x|<|y|
+		.byte %11000000		;; x+,y-,|x|>|y|
+		.byte %11111111		;; x+,y-,|x|<|y|
+		.byte %01000000		;; x-,y+,|x|>|y|
+		.byte %01111111		;; x-,y+,|x|<|y|
+		.byte %10111111		;; x-,y-,|x|>|y|
+		.byte %10000000		;; x-,y-,|x|<|y|
+
+
+		;;;;;;;; atan(2^(x/32))*128/pi ;;;;;;;;
+
+atan_tab
+		.byte $00,$00,$00,$00,$00,$00,$00,$00
+		.byte $00,$00,$00,$00,$00,$00,$00,$00
+		.byte $00,$00,$00,$00,$00,$00,$00,$00
+		.byte $00,$00,$00,$00,$00,$00,$00,$00
+		.byte $00,$00,$00,$00,$00,$00,$00,$00
+		.byte $00,$00,$00,$00,$00,$00,$00,$00
+		.byte $00,$00,$00,$00,$00,$00,$00,$00
+		.byte $00,$00,$00,$00,$00,$00,$00,$00
+		.byte $00,$00,$00,$00,$00,$00,$00,$00
+		.byte $00,$00,$00,$00,$00,$00,$00,$00
+		.byte $00,$00,$00,$00,$00,$01,$01,$01
+		.byte $01,$01,$01,$01,$01,$01,$01,$01
+		.byte $01,$01,$01,$01,$01,$01,$01,$01
+		.byte $01,$01,$01,$01,$01,$01,$01,$01
+		.byte $01,$01,$01,$01,$01,$02,$02,$02
+		.byte $02,$02,$02,$02,$02,$02,$02,$02
+		.byte $02,$02,$02,$02,$02,$02,$02,$02
+		.byte $03,$03,$03,$03,$03,$03,$03,$03
+		.byte $03,$03,$03,$03,$03,$04,$04,$04
+		.byte $04,$04,$04,$04,$04,$04,$04,$04
+		.byte $05,$05,$05,$05,$05,$05,$05,$05
+		.byte $06,$06,$06,$06,$06,$06,$06,$06
+		.byte $07,$07,$07,$07,$07,$07,$08,$08
+		.byte $08,$08,$08,$08,$09,$09,$09,$09
+		.byte $09,$0a,$0a,$0a,$0a,$0b,$0b,$0b
+		.byte $0b,$0c,$0c,$0c,$0c,$0d,$0d,$0d
+		.byte $0d,$0e,$0e,$0e,$0e,$0f,$0f,$0f
+		.byte $10,$10,$10,$11,$11,$11,$12,$12
+		.byte $12,$13,$13,$13,$14,$14,$15,$15
+		.byte $15,$16,$16,$17,$17,$17,$18,$18
+		.byte $19,$19,$19,$1a,$1a,$1b,$1b,$1c
+		.byte $1c,$1c,$1d,$1d,$1e,$1e,$1f,$1f
+
+
+		;;;;;;;; log2(x)*32 ;;;;;;;;
+
+log2_tab
+		.byte $00,$00,$20,$32,$40,$4a,$52,$59
+		.byte $60,$65,$6a,$6e,$72,$76,$79,$7d
+		.byte $80,$82,$85,$87,$8a,$8c,$8e,$90
+		.byte $92,$94,$96,$98,$99,$9b,$9d,$9e
+		.byte $a0,$a1,$a2,$a4,$a5,$a6,$a7,$a9
+		.byte $aa,$ab,$ac,$ad,$ae,$af,$b0,$b1
+		.byte $b2,$b3,$b4,$b5,$b6,$b7,$b8,$b9
+		.byte $b9,$ba,$bb,$bc,$bd,$bd,$be,$bf
+		.byte $c0,$c0,$c1,$c2,$c2,$c3,$c4,$c4
+		.byte $c5,$c6,$c6,$c7,$c7,$c8,$c9,$c9
+		.byte $ca,$ca,$cb,$cc,$cc,$cd,$cd,$ce
+		.byte $ce,$cf,$cf,$d0,$d0,$d1,$d1,$d2
+		.byte $d2,$d3,$d3,$d4,$d4,$d5,$d5,$d5
+		.byte $d6,$d6,$d7,$d7,$d8,$d8,$d9,$d9
+		.byte $d9,$da,$da,$db,$db,$db,$dc,$dc
+		.byte $dd,$dd,$dd,$de,$de,$de,$df,$df
+		.byte $df,$e0,$e0,$e1,$e1,$e1,$e2,$e2
+		.byte $e2,$e3,$e3,$e3,$e4,$e4,$e4,$e5
+		.byte $e5,$e5,$e6,$e6,$e6,$e7,$e7,$e7
+		.byte $e7,$e8,$e8,$e8,$e9,$e9,$e9,$ea
+		.byte $ea,$ea,$ea,$eb,$eb,$eb,$ec,$ec
+		.byte $ec,$ec,$ed,$ed,$ed,$ed,$ee,$ee
+		.byte $ee,$ee,$ef,$ef,$ef,$ef,$f0,$f0
+		.byte $f0,$f1,$f1,$f1,$f1,$f1,$f2,$f2
+		.byte $f2,$f2,$f3,$f3,$f3,$f3,$f4,$f4
+		.byte $f4,$f4,$f5,$f5,$f5,$f5,$f5,$f6
+		.byte $f6,$f6,$f6,$f7,$f7,$f7,$f7,$f7
+		.byte $f8,$f8,$f8,$f8,$f9,$f9,$f9,$f9
+		.byte $f9,$fa,$fa,$fa,$fa,$fa,$fb,$fb
+		.byte $fb,$fb,$fb,$fc,$fc,$fc,$fc,$fc
+		.byte $fd,$fd,$fd,$fd,$fd,$fd,$fe,$fe
+		.byte $fe,$fe,$fe,$ff,$ff,$ff,$ff,$ff
+
+    }}
+}
+
 }

compiler/res/prog8lib/pet32/syslib.p8

@@ -0,0 +1,452 @@
+; Prog8 definitions for the Commodore PET
+; Including memory registers, I/O registers, Basic and Kernal subroutines.
+; see: https://www.pagetable.com/?p=926  ,  http://www.zimmers.net/cbmpics/cbm/PETx/petmem.txt
+
+cbm {
+    ; Commodore (CBM) common variables, vectors and kernal routines
+    %option no_symbol_prefixing
+
+        &ubyte  TIME_HI         = $8d       ; software jiffy clock, hi byte
+        &ubyte  TIME_MID        = $8e       ;  .. mid byte
+        &ubyte  TIME_LO         = $8f       ;    .. lo byte. Updated by IRQ every 1/60 sec
+        &ubyte  STATUS          = $96       ; kernal status variable for I/O
+
+        &uword  CINV            = $0090     ; IRQ vector (in ram)
+        &uword  CBINV           = $0092     ; BRK vector (in ram)
+        &uword  NMINV           = $0094     ; NMI vector (in ram)
+
+        &uword  NMI_VEC         = $FFFA     ; 6502 nmi vector, determined by the kernal if banked in
+        &uword  RESET_VEC       = $FFFC     ; 6502 reset vector, determined by the kernal if banked in
+        &uword  IRQ_VEC         = $FFFE     ; 6502 interrupt vector, determined by the kernal if banked in
+
+        ; the default addresses for the character screen chars and colors
+        const  uword  Screen    = $8000     ; to have this as an array[40*25] the compiler would have to support array size > 255
+
+
+romsub $FFC6 = CHKIN(ubyte logical @ X) clobbers(A,X) -> bool @Pc    ; define an input channel
+romsub $FFC9 = CHKOUT(ubyte logical @ X) clobbers(A,X)          ; define an output channel
+romsub $FFCC = CLRCHN() clobbers(A,X)                           ; restore default devices
+romsub $FFCF = CHRIN() clobbers(X, Y) -> ubyte @ A              ; input a character (for keyboard, read a whole line from the screen) A=byte read.
+romsub $FFD2 = CHROUT(ubyte character @ A)                           ; output a character
+romsub $FFE1 = STOP() clobbers(X) -> bool @ Pz, ubyte @ A       ; check the STOP key (and some others in A)
+romsub $FFE4 = GETIN() clobbers(X,Y) -> bool @Pc, ubyte @ A     ; get a character
+romsub $FFE7 = CLALL() clobbers(A,X)                            ; close all files
+romsub $FFEA = UDTIM() clobbers(A,X)                            ; update the software clock
+
+asmsub STOP2() clobbers(X) -> ubyte @A  {
+    ; -- check if STOP key was pressed, returns true if so.  More convenient to use than STOP() because that only sets the carry status flag.
+    %asm {{
+        jsr  cbm.STOP
+        beq  +
+        lda  #0
+        rts
++       lda  #1
+        rts
+    }}
+}
+
+asmsub SETTIM(ubyte low @ A, ubyte middle @ X, ubyte high @ Y) {
+    ; PET stub to set the software clock
+    %asm {{
+        sty  TIME_HI
+        stx  TIME_MID
+        sta  TIME_LO
+        rts
+    }}
+}
+
+asmsub RDTIM() -> ubyte @ A, ubyte @ X, ubyte @ Y {
+    ; PET stub to read the software clock (A=lo,X=mid,Y=high)
+    %asm {{
+        ldy  TIME_HI
+        ldx  TIME_MID
+        lda  TIME_LO
+        rts
+    }}
+}
+
+asmsub RDTIM16() clobbers(X) -> uword @AY {
+    ; --  like RDTIM() but only returning the lower 16 bits in AY for convenience
+    %asm {{
+        lda  TIME_LO
+        ldy  TIME_MID
+        rts
+    }}
+}
+
+
+}
+
+sys {
+    ; ------- lowlevel system routines --------
+    %option no_symbol_prefixing
+
+    const ubyte target = 32         ;  compilation target specifier.  64 = C64, 128 = C128,  16 = CommanderX16, 32=PET
+
+asmsub  init_system()  {
+    ; Initializes the machine to a sane starting state.
+    ; Called automatically by the loader program logic.
+    ; Uppercase charset is activated.
+    %asm {{
+        sei
+        cld
+        lda  #142
+        jsr  cbm.CHROUT     ; uppercase
+        lda  #147
+        jsr  cbm.CHROUT     ; clear screen
+        clc
+        clv
+        cli
+        rts
+    }}
+}
+
+asmsub  init_system_phase2()  {
+    %asm {{
+        rts     ; no phase 2 steps on the PET
+    }}
+}
+
+asmsub  cleanup_at_exit() {
+    ; executed when the main subroutine does rts
+    %asm {{
+        rts
+    }}
+}
+
+    asmsub  reset_system()  {
+        ; Soft-reset the system back to initial power-on Basic prompt.
+        %asm {{
+            sei
+            jmp  (cbm.RESET_VEC)
+        }}
+    }
+
+    asmsub waitvsync() clobbers(A) {
+        ; --- busy wait till the next vsync has occurred (approximately), without depending on custom irq handling.
+        ;     Note: on PET this simply waits until the next jiffy clock update, I don't know if a true vsync is possible there
+        %asm {{
+            lda  #1
+            ldy  #0
+            jmp  wait
+        }}
+    }
+
+    asmsub wait(uword jiffies @AY) {
+        ; --- wait approximately the given number of jiffies (1/60th seconds) (N or N+1)
+        ;     note: the system irq handler has to be active for this to work as it depends on the system jiffy clock
+        %asm {{
+            stx  P8ZP_SCRATCH_B1
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+
+_loop       lda  P8ZP_SCRATCH_W1
+            ora  P8ZP_SCRATCH_W1+1
+            bne  +
+            ldx  P8ZP_SCRATCH_B1
+            rts
+
++           lda  cbm.TIME_LO
+            sta  P8ZP_SCRATCH_B1
+-           lda  cbm.TIME_LO
+            cmp  P8ZP_SCRATCH_B1
+            beq  -
+
+            lda  P8ZP_SCRATCH_W1
+            bne  +
+            dec  P8ZP_SCRATCH_W1+1
++           dec  P8ZP_SCRATCH_W1
+            jmp  _loop
+        }}
+    }
+
+    asmsub internal_stringcopy(uword source @R0, uword target @AY) clobbers (A,Y) {
+        ; Called when the compiler wants to assign a string value to another string.
+        %asm {{
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		lda  cx16.r0
+		ldy  cx16.r0+1
+		jmp  prog8_lib.strcpy
+        }}
+    }
+
+    asmsub memcopy(uword source @R0, uword target @R1, uword count @AY) clobbers(A,X,Y) {
+        ; note: only works for NON-OVERLAPPING memory regions!
+        ; note: can't be inlined because is called from asm as well
+        %asm {{
+            ldx  cx16.r0
+            stx  P8ZP_SCRATCH_W1        ; source in ZP
+            ldx  cx16.r0+1
+            stx  P8ZP_SCRATCH_W1+1
+            ldx  cx16.r1
+            stx  P8ZP_SCRATCH_W2        ; target in ZP
+            ldx  cx16.r1+1
+            stx  P8ZP_SCRATCH_W2+1
+            cpy  #0
+            bne  _longcopy
+
+            ; copy <= 255 bytes
+            tay
+            bne  _copyshort
+            rts     ; nothing to copy
+
+_copyshort
+            ; decrease source and target pointers so we can simply index by Y
+            lda  P8ZP_SCRATCH_W1
+            bne  +
+            dec  P8ZP_SCRATCH_W1+1
++           dec  P8ZP_SCRATCH_W1
+            lda  P8ZP_SCRATCH_W2
+            bne  +
+            dec  P8ZP_SCRATCH_W2+1
++           dec  P8ZP_SCRATCH_W2
+
+-           lda  (P8ZP_SCRATCH_W1),y
+            sta  (P8ZP_SCRATCH_W2),y
+            dey
+            bne  -
+            rts
+
+_longcopy
+            sta  P8ZP_SCRATCH_B1        ; lsb(count) = remainder in last page
+            tya
+            tax                         ; x = num pages (1+)
+            ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            sta  (P8ZP_SCRATCH_W2),y
+            iny
+            bne  -
+            inc  P8ZP_SCRATCH_W1+1
+            inc  P8ZP_SCRATCH_W2+1
+            dex
+            bne  -
+            ldy  P8ZP_SCRATCH_B1
+            bne  _copyshort
+            rts
+        }}
+    }
+
+    asmsub memset(uword mem @R0, uword numbytes @R1, ubyte value @A) clobbers(A,X,Y) {
+        %asm {{
+            ldy  cx16.r0
+            sty  P8ZP_SCRATCH_W1
+            ldy  cx16.r0+1
+            sty  P8ZP_SCRATCH_W1+1
+            ldx  cx16.r1
+            ldy  cx16.r1+1
+            jmp  prog8_lib.memset
+        }}
+    }
+
+    asmsub memsetw(uword mem @R0, uword numwords @R1, uword value @AY) clobbers(A,X,Y) {
+        %asm {{
+            ldx  cx16.r0
+            stx  P8ZP_SCRATCH_W1
+            ldx  cx16.r0+1
+            stx  P8ZP_SCRATCH_W1+1
+            ldx  cx16.r1
+            stx  P8ZP_SCRATCH_W2
+            ldx  cx16.r1+1
+            stx  P8ZP_SCRATCH_W2+1
+            jmp  prog8_lib.memsetw
+        }}
+    }
+
+    inline asmsub read_flags() -> ubyte @A {
+        %asm {{
+        php
+        pla
+        }}
+    }
+
+    inline asmsub clear_carry() {
+        %asm {{
+        clc
+        }}
+    }
+
+    inline asmsub set_carry() {
+        %asm {{
+        sec
+        }}
+    }
+
+    inline asmsub clear_irqd() {
+        %asm {{
+        cli
+        }}
+    }
+
+    inline asmsub set_irqd() {
+        %asm {{
+        sei
+        }}
+    }
+
+    inline asmsub irqsafe_set_irqd() {
+        %asm {{
+        php
+        sei
+        }}
+    }
+
+    inline asmsub irqsafe_clear_irqd() {
+        %asm {{
+        plp
+        }}
+    }
+
+    inline asmsub exit(ubyte returnvalue @A) {
+        ; -- immediately exit the program with a return code in the A register
+        %asm {{
+            ldx  prog8_lib.orig_stackpointer
+            txs
+            rts		; return to original caller
+        }}
+    }
+
+    inline asmsub progend() -> uword @AY {
+        %asm {{
+            lda  #<prog8_program_end
+            ldy  #>prog8_program_end
+        }}
+    }
+
+}
+
+cx16 {
+    %option no_symbol_prefixing
+
+    ; the sixteen virtual 16-bit registers that the CX16 has defined in the zeropage
+    ; they are simulated on the PET as well but their location in memory is different
+    ; (because there's no room for them in the zeropage)
+    ; we select the top page of RAM (assume 32Kb)
+    &uword r0  = $7fe0
+    &uword r1  = $7fe2
+    &uword r2  = $7fe4
+    &uword r3  = $7fe6
+    &uword r4  = $7fe8
+    &uword r5  = $7fea
+    &uword r6  = $7fec
+    &uword r7  = $7fee
+    &uword r8  = $7ff0
+    &uword r9  = $7ff2
+    &uword r10 = $7ff4
+    &uword r11 = $7ff6
+    &uword r12 = $7ff8
+    &uword r13 = $7ffa
+    &uword r14 = $7ffc
+    &uword r15 = $7ffe
+
+    &word r0s  = $7fe0
+    &word r1s  = $7fe2
+    &word r2s  = $7fe4
+    &word r3s  = $7fe6
+    &word r4s  = $7fe8
+    &word r5s  = $7fea
+    &word r6s  = $7fec
+    &word r7s  = $7fee
+    &word r8s  = $7ff0
+    &word r9s  = $7ff2
+    &word r10s = $7ff4
+    &word r11s = $7ff6
+    &word r12s = $7ff8
+    &word r13s = $7ffa
+    &word r14s = $7ffc
+    &word r15s = $7ffe
+
+    &ubyte r0L  = $7fe0
+    &ubyte r1L  = $7fe2
+    &ubyte r2L  = $7fe4
+    &ubyte r3L  = $7fe6
+    &ubyte r4L  = $7fe8
+    &ubyte r5L  = $7fea
+    &ubyte r6L  = $7fec
+    &ubyte r7L  = $7fee
+    &ubyte r8L  = $7ff0
+    &ubyte r9L  = $7ff2
+    &ubyte r10L = $7ff4
+    &ubyte r11L = $7ff6
+    &ubyte r12L = $7ff8
+    &ubyte r13L = $7ffa
+    &ubyte r14L = $7ffc
+    &ubyte r15L = $7ffe
+
+    &ubyte r0H  = $7fe1
+    &ubyte r1H  = $7fe3
+    &ubyte r2H  = $7fe5
+    &ubyte r3H  = $7fe7
+    &ubyte r4H  = $7fe9
+    &ubyte r5H  = $7feb
+    &ubyte r6H  = $7fed
+    &ubyte r7H  = $7fef
+    &ubyte r8H  = $7ff1
+    &ubyte r9H  = $7ff3
+    &ubyte r10H = $7ff5
+    &ubyte r11H = $7ff7
+    &ubyte r12H = $7ff9
+    &ubyte r13H = $7ffb
+    &ubyte r14H = $7ffd
+    &ubyte r15H = $7fff
+
+    &byte r0sL  = $7fe0
+    &byte r1sL  = $7fe2
+    &byte r2sL  = $7fe4
+    &byte r3sL  = $7fe6
+    &byte r4sL  = $7fe8
+    &byte r5sL  = $7fea
+    &byte r6sL  = $7fec
+    &byte r7sL  = $7fee
+    &byte r8sL  = $7ff0
+    &byte r9sL  = $7ff2
+    &byte r10sL = $7ff4
+    &byte r11sL = $7ff6
+    &byte r12sL = $7ff8
+    &byte r13sL = $7ffa
+    &byte r14sL = $7ffc
+    &byte r15sL = $7ffe
+
+    &byte r0sH  = $7fe1
+    &byte r1sH  = $7fe3
+    &byte r2sH  = $7fe5
+    &byte r3sH  = $7fe7
+    &byte r4sH  = $7fe9
+    &byte r5sH  = $7feb
+    &byte r6sH  = $7fed
+    &byte r7sH  = $7fef
+    &byte r8sH  = $7ff1
+    &byte r9sH  = $7ff3
+    &byte r10sH = $7ff5
+    &byte r11sH = $7ff7
+    &byte r12sH = $7ff9
+    &byte r13sH = $7ffb
+    &byte r14sH = $7ffd
+    &byte r15sH = $7fff
+
+    asmsub save_virtual_registers() clobbers(A,Y) {
+        %asm {{
+            ldy  #31
+    -       lda  cx16.r0,y
+            sta  _cx16_vreg_storage,y
+            dey
+            bpl  -
+            rts
+
+    _cx16_vreg_storage
+            .word 0,0,0,0,0,0,0,0
+            .word 0,0,0,0,0,0,0,0
+        }}
+    }
+
+    asmsub restore_virtual_registers() clobbers(A,Y) {
+        %asm {{
+            ldy  #31
+    -       lda  save_virtual_registers._cx16_vreg_storage,y
+            sta  cx16.r0,y
+            dey
+            bpl  -
+            rts
+        }}
+    }
+}

compiler/res/prog8lib/pet32/textio.p8

@@ -0,0 +1,456 @@
+; Prog8 definitions for the Text I/O and Screen routines for the Commodore PET
+
+%import syslib
+%import conv
+
+
+txt {
+    %option no_symbol_prefixing
+
+const ubyte DEFAULT_WIDTH = 40
+const ubyte DEFAULT_HEIGHT = 25
+
+
+sub  clear_screen() {
+    txt.chrout(147)
+}
+
+sub home() {
+    txt.chrout(19)
+}
+
+sub nl() {
+    txt.chrout('\n')
+}
+
+sub spc() {
+    txt.chrout(' ')
+}
+
+
+asmsub  fill_screen (ubyte character @ A, ubyte color @ Y) clobbers(A)  {
+	; ---- fill the character screen with the given fill character. color is ignored on PET
+	%asm {{
+		jmp  clear_screenchars
+        }}
+
+}
+
+asmsub  clear_screenchars (ubyte character @ A) clobbers(Y)  {
+	; ---- clear the character screen with the given fill character
+	;      (assumes screen matrix is at the default address)
+	%asm {{
+		ldy  #250
+-		sta  cbm.Screen+250*0-1,y
+		sta  cbm.Screen+250*1-1,y
+		sta  cbm.Screen+250*2-1,y
+		sta  cbm.Screen+250*3-1,y
+		dey
+		bne  -
+		rts
+        }}
+}
+
+sub  clear_screencolors (ubyte color)  {
+    ; --- dummy function on PET
+}
+
+sub color (ubyte txtcol) {
+    ; --- dummy function on PET
+}
+
+
+sub lowercase() {
+    txt.chrout(14)
+}
+
+sub uppercase() {
+    txt.chrout(142)
+}
+
+asmsub  scroll_left  () clobbers(A, X, Y)  {
+	; ---- scroll the whole screen 1 character to the left
+	;      contents of the rightmost column are unchanged, you should clear/refill this yourself
+
+	%asm {{
+		ldx  #0
+		ldy  #38
+-
+        .for row=0, row<=24, row+=1
+            lda  cbm.Screen + 40*row + 1,x
+            sta  cbm.Screen + 40*row + 0,x
+        .next
+		inx
+		dey
+		bpl  -
+
+		rts
+	}}
+}
+
+asmsub  scroll_right  () clobbers(A,X)  {
+	; ---- scroll the whole screen 1 character to the right
+	;      contents of the leftmost column are unchanged, you should clear/refill this yourself
+	%asm {{
+		ldx  #38
+-
+        .for row=0, row<=24, row+=1
+            lda  cbm.Screen + 40*row + 0,x
+            sta  cbm.Screen + 40*row + 1,x
+        .next
+		dex
+		bpl  -
+
+		rts
+	}}
+}
+
+asmsub  scroll_up  () clobbers(A,X)  {
+	; ---- scroll the whole screen 1 character up
+	;      contents of the bottom row are unchanged, you should refill/clear this yourself
+	%asm {{
+		ldx #39
+-
+        .for row=1, row<=24, row+=1
+            lda  cbm.Screen + 40*row,x
+            sta  cbm.Screen + 40*(row-1),x
+        .next
+		dex
+		bpl  -
+
+		rts
+	}}
+}
+
+asmsub  scroll_down  () clobbers(A,X)  {
+	; ---- scroll the whole screen 1 character down
+	;      contents of the top row are unchanged, you should refill/clear this yourself
+	%asm {{
+		ldx #39
+-
+        .for row=23, row>=0, row-=1
+            lda  cbm.Screen + 40*row,x
+            sta  cbm.Screen + 40*(row+1),x
+        .next
+		dex
+		bpl  -
+
+		rts
+	}}
+}
+
+romsub $FFD2 = chrout(ubyte character @ A)    ; for consistency. You can also use cbm.CHROUT directly ofcourse.
+
+asmsub  print (str text @ AY) clobbers(A,Y)  {
+	; ---- print null terminated string from A/Y
+	; note: the compiler contains an optimization that will replace
+	;       a call to this subroutine with a string argument of just one char,
+	;       by just one call to cbm.CHROUT of that single char.
+	%asm {{
+		sta  P8ZP_SCRATCH_B1
+		sty  P8ZP_SCRATCH_REG
+		ldy  #0
+-		lda  (P8ZP_SCRATCH_B1),y
+		beq  +
+		jsr  cbm.CHROUT
+		iny
+		bne  -
++		rts
+	}}
+}
+
+asmsub  print_ub0  (ubyte value @ A) clobbers(A,X,Y)  {
+	; ---- print the ubyte in A in decimal form, with left padding 0s (3 positions total)
+	%asm {{
+		jsr  conv.ubyte2decimal
+		pha
+		tya
+		jsr  cbm.CHROUT
+		pla
+		jsr  cbm.CHROUT
+		txa
+		jmp  cbm.CHROUT
+	}}
+}
+
+asmsub  print_ub  (ubyte value @ A) clobbers(A,X,Y)  {
+	; ---- print the ubyte in A in decimal form, without left padding 0s
+	%asm {{
+		jsr  conv.ubyte2decimal
+_print_byte_digits
+		pha
+		cpy  #'0'
+		beq  +
+		tya
+		jsr  cbm.CHROUT
+		pla
+		jsr  cbm.CHROUT
+		jmp  _ones
++       pla
+        cmp  #'0'
+        beq  _ones
+        jsr  cbm.CHROUT
+_ones   txa
+		jmp  cbm.CHROUT
+	}}
+}
+
+asmsub  print_b  (byte value @ A) clobbers(A,X,Y)  {
+	; ---- print the byte in A in decimal form, without left padding 0s
+	%asm {{
+		pha
+		cmp  #0
+		bpl  +
+		lda  #'-'
+		jsr  cbm.CHROUT
++		pla
+		jsr  conv.byte2decimal
+		jmp  print_ub._print_byte_digits
+	}}
+}
+
+asmsub  print_ubhex  (ubyte value @ A, bool prefix @ Pc) clobbers(A,X,Y)  {
+	; ---- print the ubyte in A in hex form (if Carry is set, a radix prefix '$' is printed as well)
+	%asm {{
+		bcc  +
+		pha
+		lda  #'$'
+		jsr  cbm.CHROUT
+		pla
++		jsr  conv.ubyte2hex
+		jsr  cbm.CHROUT
+		tya
+		jmp  cbm.CHROUT
+	}}
+}
+
+asmsub  print_ubbin  (ubyte value @ A, bool prefix @ Pc) clobbers(A,X,Y)  {
+	; ---- print the ubyte in A in binary form (if Carry is set, a radix prefix '%' is printed as well)
+	%asm {{
+		sta  P8ZP_SCRATCH_B1
+		bcc  +
+		lda  #'%'
+		jsr  cbm.CHROUT
++		ldy  #8
+-		lda  #'0'
+		asl  P8ZP_SCRATCH_B1
+		bcc  +
+		lda  #'1'
++		jsr  cbm.CHROUT
+		dey
+		bne  -
+		rts
+	}}
+}
+
+asmsub  print_uwbin  (uword value @ AY, bool prefix @ Pc) clobbers(A,X,Y)  {
+	; ---- print the uword in A/Y in binary form (if Carry is set, a radix prefix '%' is printed as well)
+	%asm {{
+		pha
+		tya
+		jsr  print_ubbin
+		pla
+		clc
+		jmp  print_ubbin
+	}}
+}
+
+asmsub  print_uwhex  (uword value @ AY, bool prefix @ Pc) clobbers(A,X,Y)  {
+	; ---- print the uword in A/Y in hexadecimal form (4 digits)
+	;      (if Carry is set, a radix prefix '$' is printed as well)
+	%asm {{
+		pha
+		tya
+		jsr  print_ubhex
+		pla
+		clc
+		jmp  print_ubhex
+	}}
+}
+
+asmsub  print_uw0  (uword value @ AY) clobbers(A,X,Y)  {
+	; ---- print the uword in A/Y in decimal form, with left padding 0s (5 positions total)
+	%asm {{
+		jsr  conv.uword2decimal
+		ldy  #0
+-		lda  conv.uword2decimal.decTenThousands,y
+        beq  +
+		jsr  cbm.CHROUT
+		iny
+		bne  -
++		rts
+	}}
+}
+
+asmsub  print_uw  (uword value @ AY) clobbers(A,X,Y)  {
+	; ---- print the uword in A/Y in decimal form, without left padding 0s
+	%asm {{
+		jsr  conv.uword2decimal
+		ldy  #0
+-		lda  conv.uword2decimal.decTenThousands,y
+		beq  _allzero
+		cmp  #'0'
+		bne  _gotdigit
+		iny
+		bne  -
+
+_gotdigit
+		jsr  cbm.CHROUT
+		iny
+		lda  conv.uword2decimal.decTenThousands,y
+		bne  _gotdigit
+		rts
+_allzero
+        lda  #'0'
+        jmp  cbm.CHROUT
+	}}
+}
+
+asmsub  print_w  (word value @ AY) clobbers(A,X,Y)  {
+	; ---- print the (signed) word in A/Y in decimal form, without left padding 0's
+	%asm {{
+		cpy  #0
+		bpl  +
+		pha
+		lda  #'-'
+		jsr  cbm.CHROUT
+		tya
+		eor  #255
+		tay
+		pla
+		eor  #255
+		clc
+		adc  #1
+		bcc  +
+		iny
++		jmp  print_uw
+	}}
+}
+
+asmsub  input_chars  (uword buffer @ AY) clobbers(A) -> ubyte @ Y  {
+	; ---- Input a string (max. 80 chars) from the keyboard. Returns length in Y. (string is terminated with a 0 byte as well)
+	;      It assumes the keyboard is selected as I/O channel!
+
+	%asm {{
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		ldy  #0				; char counter = 0
+-		jsr  cbm.CHRIN
+		cmp  #$0d			; return (ascii 13) pressed?
+		beq  +				; yes, end.
+		sta  (P8ZP_SCRATCH_W1),y	; else store char in buffer
+		iny
+		bne  -
++		lda  #0
+		sta  (P8ZP_SCRATCH_W1),y	; finish string with 0 byte
+		rts
+
+	}}
+}
+
+asmsub  setchr  (ubyte col @X, ubyte row @Y, ubyte character @A) clobbers(A, Y)  {
+	; ---- sets the character in the screen matrix at the given position
+	%asm {{
+		pha
+		tya
+		asl  a
+		tay
+		lda  _screenrows+1,y
+		sta  _mod+2
+		txa
+		clc
+		adc  _screenrows,y
+		sta  _mod+1
+		bcc  +
+		inc  _mod+2
++		pla
+_mod		sta  $ffff		; modified
+		rts
+
+_screenrows	.word  cbm.Screen + range(0, 1000, 40)
+	}}
+}
+
+asmsub  getchr  (ubyte col @A, ubyte row @Y) clobbers(Y) -> ubyte @ A {
+	; ---- get the character in the screen matrix at the given location
+	%asm  {{
+		pha
+		tya
+		asl  a
+		tay
+		lda  setchr._screenrows+1,y
+		sta  _mod+2
+		pla
+		clc
+		adc  setchr._screenrows,y
+		sta  _mod+1
+		bcc  _mod
+		inc  _mod+2
+_mod		lda  $ffff		; modified
+		rts
+	}}
+}
+
+sub  setclr  (ubyte col, ubyte row, ubyte color)  {
+    ; --- dummy function on PET
+}
+
+
+sub  setcc  (ubyte col, ubyte row, ubyte character, ubyte charcolor)  {
+	; ---- set char at the given position on the screen. charcolor is ignored on PET
+	%asm {{
+  		lda  row
+		asl  a
+		tay
+		lda  setchr._screenrows+1,y
+		sta  _charmod+2
+		lda  setchr._screenrows,y
+		clc
+		adc  col
+		sta  _charmod+1
+		bcc  +
+		inc  _charmod+2
++		lda  character
+_charmod	sta  $ffff		; modified
+		rts
+	}}
+}
+
+asmsub  plot  (ubyte col @ Y, ubyte row @ X) {
+	%asm  {{
+	    jsr  home
+	    cpy  #0
+	    beq  +
+-	    lda  #17
+	    jsr  chrout
+	    dey
+	    bne  -
++	    cpx  #0
+	    beq  +
+-       lda  #29
+        jsr  chrout
+        dex
+        bne  -
++		rts
+	}}
+}
+
+asmsub width() clobbers(X,Y) -> ubyte @A {
+    ; -- returns the text screen width (number of columns)
+    %asm {{
+        lda  $d5
+        clc
+        adc  #1
+        rts
+    }}
+}
+
+asmsub height() clobbers(X, Y) -> ubyte @A {
+    ; -- returns the text screen height (number of rows)
+    %asm {{
+        lda  #25
+        rts
+    }}
+}
+
+}

compiler/res/prog8lib/prog8_funcs.asm

@@ -1,20 +1,6 @@
 ; ---- builtin functions
 
 
-func_any_b_stack	.proc
-		jsr  func_any_b_into_A
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
-
-func_all_b_stack	.proc
-		jsr  func_all_b_into_A
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
-
 func_any_b_into_A	.proc
 		; -- any(array),  array in P8ZP_SCRATCH_W1, num bytes in A
 		sta  _cmp_mod+1		; self-modifying code
@@ -49,14 +35,6 @@ func_any_w_into_A	.proc
 		jmp  func_any_b_into_A
 		.pend
 
-func_any_w_stack	.proc
-		asl  a
-		jsr  func_any_b_into_A
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
-
 func_all_w_into_A	.proc
 		; -- all(warray),  array in P8ZP_SCRATCH_W1, num bytes in A
 		asl  a			; times 2 because of word
@@ -77,22 +55,6 @@ _cmp_mod	cpy  #255		; modified
 		rts
 		.pend
 
-func_all_w_stack	.proc
-		jsr  func_all_w_into_A
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
-
-abs_b_stack	.proc
-	; -- push abs(A) on stack (as unsigned word)
-		jsr  abs_b_into_A
-		sta  P8ESTACK_LO,x
-		stz  P8ESTACK_HI,x
-		dex
-		rts
-		.pend
-
 abs_b_into_A	.proc
 	; -- A = abs(A)
 		cmp  #0
@@ -104,16 +66,6 @@ abs_b_into_A	.proc
 		rts
 		.pend
 
-abs_w_stack	.proc
-	; -- push abs(AY) on stack (as word)
-		jsr  abs_w_into_AY
-		sta  P8ESTACK_LO,x
-		tya
-		sta  P8ESTACK_HI,x
-		dex
-		rts
-		.pend
-
 abs_w_into_AY	.proc
 	; -- AY = abs(AY)
 		cpy  #0
@@ -142,13 +94,6 @@ _neg		lda  #-1
 		rts
 		.pend
 
-func_sign_b_stack	.proc
-		jsr  func_sign_b_into_A
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
-
 func_sign_ub_into_A	.proc
 		cmp  #0
 		bne  _pos
@@ -157,13 +102,6 @@ _pos		lda  #1
 		rts
 		.pend
 
-func_sign_ub_stack	.proc
-		jsr  func_sign_ub_into_A
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
-
 func_sign_uw_into_A	.proc
 		cpy  #0
 		beq  _possibly_zero
@@ -174,13 +112,6 @@ _possibly_zero	cmp  #0
 		rts
 		.pend
 
-func_sign_uw_stack	.proc
-		jsr  func_sign_uw_into_A
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
-
 func_sign_w_into_A	.proc
 		cpy  #0
 		beq  _possibly_zero
@@ -195,52 +126,43 @@ _possibly_zero	cmp  #0
 		.pend
 
 
-func_sign_w_stack	.proc
-		jsr  func_sign_w_into_A
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
-
-func_sqrt16_stack	.proc
-		jsr  func_sqrt16_into_A
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
-
 func_sqrt16_into_A	.proc
-		; integer square root from  http://6502org.wikidot.com/software-math-sqrt
-		sta  P8ZP_SCRATCH_W1
-		sty  P8ZP_SCRATCH_W1+1
-		txa
-		pha
-		lda  #0
-		sta  P8ZP_SCRATCH_B1
-		sta  P8ZP_SCRATCH_REG
-		ldx  #8
--		sec
-		lda  P8ZP_SCRATCH_W1+1
-		sbc  #$40
-		tay
-		lda  P8ZP_SCRATCH_REG
-		sbc  P8ZP_SCRATCH_B1
-		bcc  +
-		sty  P8ZP_SCRATCH_W1+1
-		sta  P8ZP_SCRATCH_REG
-+		rol  P8ZP_SCRATCH_B1
-		asl  P8ZP_SCRATCH_W1
-		rol  P8ZP_SCRATCH_W1+1
-		rol  P8ZP_SCRATCH_REG
-		asl  P8ZP_SCRATCH_W1
-		rol  P8ZP_SCRATCH_W1+1
-		rol  P8ZP_SCRATCH_REG
-		dex
-		bne  -
-		pla
-		tax
-		lda  P8ZP_SCRATCH_B1
-		rts
+		; integer square root
+		; http://6502org.wikidot.com/software-math-sqrt
+		; https://github.com/TobyLobster/sqrt_test/blob/main/sqrt/sqrt7.a
+		; Tweaked by TobyLobster and 0xC0DE to be smaller and faster
+_numl = P8ZP_SCRATCH_W1
+_numh = P8ZP_SCRATCH_W1+1
+_loop_counter = P8ZP_SCRATCH_REG
+_root = P8ZP_SCRATCH_B1
+            sta  _numl
+            sty  _numh
+            ldx  #$ff
+            stx  _loop_counter
+            inx
+            stx  _root
+            sec
+_loop       lda  _numh
+            sbc  #$40
+            tay
+            txa
+            sbc  _root
+            bcc  +
+            sty  _numh
+            bcs  ++
++           txa
++           rol  _root
+            asl  _numl
+            rol  _numh
+            rol  a
+            asl  _numl
+            rol  _numh
+            rol  a
+            tax
+            lsr  _loop_counter
+            bne  _loop
+            lda  _root
+            rts
 		.pend
 
 

compiler/res/prog8lib/prog8_lib.asm

@@ -4,10 +4,8 @@
 
 orig_stackpointer	.byte  0	; stores the Stack pointer register at program start
 
-read_byte_from_address_on_stack	.proc
-	; -- read the byte from the memory address on the top of the stack, return in A (stack remains unchanged)
-		lda  P8ESTACK_LO+1,x
-		ldy  P8ESTACK_HI+1,x
+
+read_byte_from_address_in_AY_into_A	.proc
 		sta  P8ZP_SCRATCH_W2
 		sty  P8ZP_SCRATCH_W2+1
 		ldy  #0
@@ -16,319 +14,16 @@ read_byte_from_address_on_stack	.proc
 		.pend
 
 
-write_byte_to_address_on_stack	.proc
-	; -- write the byte in A to the memory address on the top of the stack (stack remains unchanged)
-		ldy  P8ESTACK_LO+1,x
-		sty  P8ZP_SCRATCH_W2
-		ldy  P8ESTACK_HI+1,x
+write_byte_X_to_address_in_AY	.proc
+		sta  P8ZP_SCRATCH_W2
 		sty  P8ZP_SCRATCH_W2+1
 		ldy  #0
+		txa
 		sta  (P8ZP_SCRATCH_W2),y
 		rts
 		.pend
 
 
-
-neg_b		.proc
-		lda  #0
-		sec
-		sbc  P8ESTACK_LO+1,x
-		sta  P8ESTACK_LO+1,x
-		rts
-		.pend
-
-neg_w		.proc
-		sec
-		lda  #0
-		sbc  P8ESTACK_LO+1,x
-		sta  P8ESTACK_LO+1,x
-		lda  #0
-		sbc  P8ESTACK_HI+1,x
-		sta  P8ESTACK_HI+1,x
-		rts
-		.pend
-
-inv_word	.proc
-		lda  P8ESTACK_LO+1,x
-		eor  #255
-		sta  P8ESTACK_LO+1,x
-		lda  P8ESTACK_HI+1,x
-		eor  #255
-		sta  P8ESTACK_HI+1,x
-		rts
-		.pend
-
-bitand_b	.proc
-		; -- bitwise and (of 2 bytes)
-		lda  P8ESTACK_LO+2,x
-		and  P8ESTACK_LO+1,x
-		inx
-		sta  P8ESTACK_LO+1,x
-		rts
-		.pend
-
-bitor_b		.proc
-		; -- bitwise or (of 2 bytes)
-		lda  P8ESTACK_LO+2,x
-		ora  P8ESTACK_LO+1,x
-		inx
-		sta  P8ESTACK_LO+1,x
-		rts
-		.pend
-
-bitxor_b	.proc
-		; -- bitwise xor (of 2 bytes)
-		lda  P8ESTACK_LO+2,x
-		eor  P8ESTACK_LO+1,x
-		inx
-		sta  P8ESTACK_LO+1,x
-		rts
-		.pend
-
-bitand_w	.proc
-		; -- bitwise and (of 2 words)
-		lda  P8ESTACK_LO+2,x
-		and  P8ESTACK_LO+1,x
-		sta  P8ESTACK_LO+2,x
-		lda  P8ESTACK_HI+2,x
-		and  P8ESTACK_HI+1,x
-		sta  P8ESTACK_HI+2,x
-		inx
-		rts
-		.pend
-
-bitor_w		.proc
-		; -- bitwise or (of 2 words)
-		lda  P8ESTACK_LO+2,x
-		ora  P8ESTACK_LO+1,x
-		sta  P8ESTACK_LO+2,x
-		lda  P8ESTACK_HI+2,x
-		ora  P8ESTACK_HI+1,x
-		sta  P8ESTACK_HI+2,x
-		inx
-		rts
-		.pend
-
-bitxor_w	.proc
-		; -- bitwise xor (of 2 bytes)
-		lda  P8ESTACK_LO+2,x
-		eor  P8ESTACK_LO+1,x
-		sta  P8ESTACK_LO+2,x
-		lda  P8ESTACK_HI+2,x
-		eor  P8ESTACK_HI+1,x
-		sta  P8ESTACK_HI+2,x
-		inx
-		rts
-		.pend
-
-
-add_w		.proc
-	; -- push word+word / uword+uword
-		inx
-		clc
-		lda  P8ESTACK_LO,x
-		adc  P8ESTACK_LO+1,x
-		sta  P8ESTACK_LO+1,x
-		lda  P8ESTACK_HI,x
-		adc  P8ESTACK_HI+1,x
-		sta  P8ESTACK_HI+1,x
-		rts
-		.pend
-
-sub_w		.proc
-	; -- push word-word
-		inx
-		sec
-		lda  P8ESTACK_LO+1,x
-		sbc  P8ESTACK_LO,x
-		sta  P8ESTACK_LO+1,x
-		lda  P8ESTACK_HI+1,x
-		sbc  P8ESTACK_HI,x
-		sta  P8ESTACK_HI+1,x
-		rts
-		.pend
-
-mul_byte	.proc
-	; -- b*b->b (signed and unsigned)
-		inx
-		lda  P8ESTACK_LO,x
-		ldy  P8ESTACK_LO+1,x
-		jsr  math.multiply_bytes
-		sta  P8ESTACK_LO+1,x
-		rts
-		.pend
-
-mul_word	.proc
-		inx
-		lda  P8ESTACK_LO,x
-		sta  P8ZP_SCRATCH_W1
-		lda  P8ESTACK_HI,x
-		sta  P8ZP_SCRATCH_W1+1
-		lda  P8ESTACK_LO+1,x
-		ldy  P8ESTACK_HI+1,x
-		jsr  math.multiply_words
-		lda  math.multiply_words.result
-		sta  P8ESTACK_LO+1,x
-		lda  math.multiply_words.result+1
-		sta  P8ESTACK_HI+1,x
-		rts
-		.pend
-
-idiv_b		.proc
-	; signed division: use unsigned division and fix sign of result afterwards
-		inx
-		lda  P8ESTACK_LO,x
-		eor  P8ESTACK_LO+1,x
-		php			; save sign of result
-		lda  P8ESTACK_LO,x
-		bpl  +
-		eor  #$ff
-		sec
-		adc  #0			; make num1 positive
-+		tay
-		inx
-		lda  P8ESTACK_LO,x
-		bpl  +
-		eor  #$ff
-		sec
-		adc  #0			; make num2 positive
-+		jsr  math.divmod_ub_asm
-		sta  _remainder
-		tya
-		plp			; get sign of result
-		bpl  +
-		eor  #$ff
-		sec
-		adc  #0			; negate result
-+		sta  P8ESTACK_LO,x
-		dex
-		rts
-_remainder	.byte  0
-		.pend
-
-idiv_ub		.proc
-		inx
-		ldy  P8ESTACK_LO,x
-		lda  P8ESTACK_LO+1,x
-		jsr  math.divmod_ub_asm
-		tya
-		sta  P8ESTACK_LO+1,x
-		rts
-		.pend
-
-idiv_w		.proc
-	; signed division: use unsigned division and fix sign of result afterwards
-		lda  P8ESTACK_HI+2,x
-		eor  P8ESTACK_HI+1,x
-		php				; save sign of result
-		lda  P8ESTACK_HI+1,x
-		bpl  +
-		jsr  neg_w			; make value positive
-+		inx
-		lda  P8ESTACK_HI+1,x
-		bpl  +
-		jsr  neg_w			; make value positive
-+		lda  P8ESTACK_LO+1,x
-		sta  P8ZP_SCRATCH_W1
-		lda  P8ESTACK_HI+1,x
-		sta  P8ZP_SCRATCH_W1+1
-		lda  P8ESTACK_LO,x
-		ldy  P8ESTACK_HI,x
-		jsr  math.divmod_uw_asm
-		sta  P8ESTACK_LO+1,x
-		tya
-		sta  P8ESTACK_HI+1,x
-		plp
-		bpl  +
-		jmp  neg_w		; negate result
-+		rts
-		.pend
-
-idiv_uw		.proc
-		inx
-		lda  P8ESTACK_LO+1,x
-		sta  P8ZP_SCRATCH_W1
-		lda  P8ESTACK_HI+1,x
-		sta  P8ZP_SCRATCH_W1+1
-		lda  P8ESTACK_LO,x
-		ldy  P8ESTACK_HI,x
-		jsr  math.divmod_uw_asm
-		sta  P8ESTACK_LO+1,x
-		tya
-		sta  P8ESTACK_HI+1,x
-		rts
-		.pend
-
-remainder_ub	.proc
-		inx
-		ldy  P8ESTACK_LO,x	; right operand
-		lda  P8ESTACK_LO+1,x  ; left operand
-		jsr  math.divmod_ub_asm
-		sta  P8ESTACK_LO+1,x
-		rts
-		.pend
-
-remainder_uw	.proc
-		inx
-		lda  P8ESTACK_LO+1,x
-		sta  P8ZP_SCRATCH_W1
-		lda  P8ESTACK_HI+1,x
-		sta  P8ZP_SCRATCH_W1+1
-		lda  P8ESTACK_LO,x
-		ldy  P8ESTACK_HI,x
-		jsr  math.divmod_uw_asm
-		lda  P8ZP_SCRATCH_W2
-		sta  P8ESTACK_LO+1,x
-		lda  P8ZP_SCRATCH_W2+1
-		sta  P8ESTACK_HI+1,x
-		rts
-		.pend
-
-equal_w		.proc
-	; -- are the two words on the stack identical?
-		lda  P8ESTACK_LO+1,x
-		cmp  P8ESTACK_LO+2,x
-		bne  equal_b._equal_b_false
-		lda  P8ESTACK_HI+1,x
-		cmp  P8ESTACK_HI+2,x
-		bne  equal_b._equal_b_false
-		beq  equal_b._equal_b_true
-		.pend
-
-notequal_b	.proc
-	; -- are the two bytes on the stack different?
-		lda  P8ESTACK_LO+1,x
-		cmp  P8ESTACK_LO+2,x
-		beq  equal_b._equal_b_false
-		bne  equal_b._equal_b_true
-		.pend
-
-notequal_w	.proc
-	; -- are the two words on the stack different?
-		lda  P8ESTACK_HI+1,x
-		cmp  P8ESTACK_HI+2,x
-		beq  notequal_b
-		bne  equal_b._equal_b_true
-		.pend
-
-less_ub		.proc
-		lda  P8ESTACK_LO+2,x
-		cmp  P8ESTACK_LO+1,x
-		bcc  equal_b._equal_b_true
-		bcs  equal_b._equal_b_false
-		.pend
-
-less_b		.proc
-	; see http://www.6502.org/tutorials/compare_beyond.html
-		lda  P8ESTACK_LO+2,x
-		sec
-		sbc  P8ESTACK_LO+1,x
-		bvc  +
-		eor  #$80
-+		bmi  equal_b._equal_b_true
-		bpl  equal_b._equal_b_false
-		.pend
-
 reg_less_uw	.proc
 		;  AY < P8ZP_SCRATCH_W2?
 		cpy  P8ZP_SCRATCH_W2+1
@@ -342,17 +37,6 @@ _true		lda  #1
 		rts
 		.pend
 
-less_uw		.proc
-		lda  P8ESTACK_HI+2,x
-		cmp  P8ESTACK_HI+1,x
-		bcc  equal_b._equal_b_true
-		bne  equal_b._equal_b_false
-		lda  P8ESTACK_LO+2,x
-		cmp  P8ESTACK_LO+1,x
-		bcc  equal_b._equal_b_true
-		bcs  equal_b._equal_b_false
-		.pend
-
 reg_less_w	.proc
 		; -- AY < P8ZP_SCRATCH_W2?
 		cmp  P8ZP_SCRATCH_W2
@@ -367,48 +51,6 @@ _true		lda  #1
 		rts
 		.pend
 
-less_w		.proc
-		lda  P8ESTACK_LO+2,x
-		cmp  P8ESTACK_LO+1,x
-		lda  P8ESTACK_HI+2,x
-		sbc  P8ESTACK_HI+1,x
-		bvc  +
-		eor  #$80
-+		bmi  equal_b._equal_b_true
-		bpl  equal_b._equal_b_false
-		.pend
-
-equal_b		.proc
-	; -- are the two bytes on the stack identical?
-		lda  P8ESTACK_LO+2,x
-		cmp  P8ESTACK_LO+1,x
-		bne  _equal_b_false
-_equal_b_true	lda  #1
-_equal_b_store	inx
-		sta  P8ESTACK_LO+1,x
-		rts
-_equal_b_false	lda  #0
-		beq  _equal_b_store
-		.pend
-
-lesseq_ub	.proc
-		lda  P8ESTACK_LO+1,x
-		cmp  P8ESTACK_LO+2,x
-		bcs  equal_b._equal_b_true
-		bcc  equal_b._equal_b_false
-		.pend
-
-lesseq_b	.proc
-	; see http://www.6502.org/tutorials/compare_beyond.html
-		lda  P8ESTACK_LO+2,x
-		clc
-		sbc  P8ESTACK_LO+1,x
-		bvc  +
-		eor  #$80
-+		bmi  equal_b._equal_b_true
-		bpl  equal_b._equal_b_false
-		.pend
-
 reg_lesseq_uw	.proc
 		; AY <= P8ZP_SCRATCH_W2?
 		cpy  P8ZP_SCRATCH_W2+1
@@ -425,17 +67,6 @@ _true		lda  #1
 		rts
 		.pend
 
-lesseq_uw	.proc
-		lda  P8ESTACK_HI+1,x
-		cmp  P8ESTACK_HI+2,x
-		bcc  equal_b._equal_b_false
-		bne  equal_b._equal_b_true
-		lda  P8ESTACK_LO+1,x
-		cmp  P8ESTACK_LO+2,x
-		bcs  equal_b._equal_b_true
-		bcc  equal_b._equal_b_false
-		.pend
-
 reg_lesseq_w	.proc
 		; -- P8ZP_SCRATCH_W2 <= AY ?   (note: order different from other routines)
 		cmp  P8ZP_SCRATCH_W2
@@ -450,224 +81,6 @@ reg_lesseq_w	.proc
 		rts
 		.pend
 
-lesseq_w	.proc
-		lda  P8ESTACK_LO+1,x
-		cmp  P8ESTACK_LO+2,x
-		lda  P8ESTACK_HI+1,x
-		sbc  P8ESTACK_HI+2,x
-		bvc  +
-		eor  #$80
-+		bpl  equal_b._equal_b_true
-		bmi  equal_b._equal_b_false
-		.pend
-
-greater_ub	.proc
-		lda  P8ESTACK_LO+2,x
-		cmp  P8ESTACK_LO+1,x
-		beq  equal_b._equal_b_false
-		bcs  equal_b._equal_b_true
-		bcc  equal_b._equal_b_false
-		.pend
-
-greater_b	.proc
-	; see http://www.6502.org/tutorials/compare_beyond.html
-		lda  P8ESTACK_LO+2,x
-		clc
-		sbc  P8ESTACK_LO+1,x
-		bvc  +
-		eor  #$80
-+		bpl  equal_b._equal_b_true
-		bmi  equal_b._equal_b_false
-		.pend
-
-greater_uw	.proc
-		lda  P8ESTACK_HI+1,x
-		cmp  P8ESTACK_HI+2,x
-		bcc  equal_b._equal_b_true
-		bne  equal_b._equal_b_false
-		lda  P8ESTACK_LO+1,x
-		cmp  P8ESTACK_LO+2,x
-		bcc  equal_b._equal_b_true
-		bcs  equal_b._equal_b_false
-		.pend
-
-greater_w	.proc
-		lda  P8ESTACK_LO+1,x
-		cmp  P8ESTACK_LO+2,x
-		lda  P8ESTACK_HI+1,x
-		sbc  P8ESTACK_HI+2,x
-		bvc  +
-		eor  #$80
-+		bmi  equal_b._equal_b_true
-		bpl  equal_b._equal_b_false
-		.pend
-
-greatereq_ub	.proc
-		lda  P8ESTACK_LO+2,x
-		cmp  P8ESTACK_LO+1,x
-		bcs  equal_b._equal_b_true
-		bcc  equal_b._equal_b_false
-		.pend
-
-greatereq_b	.proc
-	; see http://www.6502.org/tutorials/compare_beyond.html
-		lda  P8ESTACK_LO+2,x
-		sec
-		sbc  P8ESTACK_LO+1,x
-		bvc  +
-		eor  #$80
-+		bpl  equal_b._equal_b_true
-		bmi  equal_b._equal_b_false
-		.pend
-
-greatereq_uw	.proc
-		lda  P8ESTACK_HI+2,x
-		cmp  P8ESTACK_HI+1,x
-		bcc  equal_b._equal_b_false
-		bne  equal_b._equal_b_true
-		lda  P8ESTACK_LO+2,x
-		cmp  P8ESTACK_LO+1,x
-		bcs  equal_b._equal_b_true
-		bcc  equal_b._equal_b_false
-		.pend
-
-greatereq_w	.proc
-		lda  P8ESTACK_LO+2,x
-		cmp  P8ESTACK_LO+1,x
-		lda  P8ESTACK_HI+2,x
-		sbc  P8ESTACK_HI+1,x
-		bvc  +
-		eor  #$80
-+		bmi  equal_b._equal_b_false
-		bpl  equal_b._equal_b_true
-		.pend
-
-
-shiftleft_b	.proc
-		inx
-		ldy  P8ESTACK_LO,x
-		bne  +
-		rts
-+		lda  P8ESTACK_LO+1,x
--		asl  a
-		dey
-		bne  -
-		sta  P8ESTACK_LO+1,x
-		rts
-		.pend
-
-shiftright_b	.proc
-		inx
-		ldy  P8ESTACK_LO,x
-		bne  +
-		rts
-+		lda  P8ESTACK_LO+1,x
--		lsr  a
-		dey
-		bne  -
-		sta  P8ESTACK_LO+1,x
-		rts
-		.pend
-
-
-equalzero_b	.proc
-		lda  P8ESTACK_LO+1,x
-		beq  _true
-		bne  _false
-_true		lda  #1
-		sta  P8ESTACK_LO+1,x
-		rts
-_false		lda  #0
-		sta  P8ESTACK_LO+1,x
-		rts
-		.pend
-
-equalzero_w	.proc
-		lda  P8ESTACK_LO+1,x
-		ora  P8ESTACK_HI+1,x
-		beq  equalzero_b._true
-		bne  equalzero_b._false
-		.pend
-
-notequalzero_b	.proc
-		lda  P8ESTACK_LO+1,x
-		beq  equalzero_b._false
-		bne  equalzero_b._true
-		.pend
-
-notequalzero_w	.proc
-		lda  P8ESTACK_LO+1,x
-		ora  P8ESTACK_HI+1,x
-		beq  equalzero_b._false
-		bne  equalzero_b._true
-		.pend
-
-lesszero_b	.proc
-		lda  P8ESTACK_LO+1,x
-		bmi  equalzero_b._true
-		jmp  equalzero_b._false
-		.pend
-
-lesszero_w	.proc
-		lda  P8ESTACK_HI+1,x
-		bmi  equalzero_b._true
-		jmp  equalzero_b._false
-		.pend
-
-greaterzero_ub	.proc
-		lda  P8ESTACK_LO+1,x
-		bne  equalzero_b._true
-		beq  equalzero_b._false
-		.pend
-
-greaterzero_sb	.proc
-		lda  P8ESTACK_LO+1,x
-		beq  equalzero_b._false
-		bpl  equalzero_b._true
-		bmi  equalzero_b._false
-		.pend
-
-greaterzero_uw	.proc
-		lda  P8ESTACK_LO+1,x
-		ora  P8ESTACK_HI+1,x
-		bne  equalzero_b._true
-		beq  equalzero_b._false
-		.pend
-
-greaterzero_sw	.proc
-		lda  P8ESTACK_HI+1,x
-		bmi  equalzero_b._false
-		ora  P8ESTACK_LO+1,x
-		beq  equalzero_b._false
-		bne  equalzero_b._true
-		.pend
-
-lessequalzero_sb	.proc
-		lda  P8ESTACK_LO+1,x
-		bmi  equalzero_b._true
-		beq  equalzero_b._true
-		bne  equalzero_b._false
-		.pend
-
-lessequalzero_sw	.proc
-		lda  P8ESTACK_HI+1,x
-		bmi  equalzero_b._true
-		ora  P8ESTACK_LO+1,x
-		beq  equalzero_b._true
-		bne  equalzero_b._false
-		.pend
-
-greaterequalzero_sb	.proc
-		lda  P8ESTACK_LO+1,x
-	    	bpl  equalzero_b._true
-	    	bmi  equalzero_b._false
-		.pend
-
-greaterequalzero_sw	.proc
-		lda  P8ESTACK_HI+1,x
-	    	bpl  equalzero_b._true
-	    	bmi  equalzero_b._false
-		.pend
 
 memcopy16_up	.proc
 	; -- copy memory UP from (P8ZP_SCRATCH_W1) to (P8ZP_SCRATCH_W2) of length X/Y (16-bit, X=lo, Y=hi)
@@ -997,21 +410,6 @@ _arg_s1		.word  0
 _arg_s2		.word  0
 		.pend
 
-strcmp_stack	.proc
-		; -- compare strings, both on stack.
-		;    Returns -1,0,1 in A, depeding on the ordering. Clobbers Y.
-		inx
-		lda  P8ESTACK_LO,x
-		ldy  P8ESTACK_HI,x
-		sta  P8ZP_SCRATCH_W2
-		sty  P8ZP_SCRATCH_W2+1
-		inx
-		lda  P8ESTACK_LO,x
-		ldy  P8ESTACK_HI,x
-		jmp  strcmp_mem
-		.pend
-
-
 strcmp_mem	.proc
 		; --   compares strings in s1 (AY) and s2 (P8ZP_SCRATCH_W2).
 		;      Returns -1,0,1 in A, depeding on the ordering. Clobbers Y.
@@ -1042,16 +440,6 @@ _return_minusone
 		.pend
 
 
-sign_extend_stack_byte	.proc
-	; -- sign extend the (signed) byte on the stack to full 16 bits
-		lda  P8ESTACK_LO+1,x
-		ora  #$7f
-		bmi  +
-		lda  #0
-+		sta  P8ESTACK_HI+1,x
-		rts
-		.pend
-
 strlen          .proc
         ; -- returns the number of bytes in the string in AY, in Y.
 		sta  P8ZP_SCRATCH_W1

compiler/res/prog8lib/prog8_lib.p8

@@ -1,6 +1,8 @@
 ; Internal library routines - always included by the compiler
 
 prog8_lib {
+    %option no_symbol_prefixing
+
 	%asminclude "library:prog8_lib.asm"
 	%asminclude "library:prog8_funcs.asm"
 }

compiler/res/prog8lib/string.p8

@@ -1,6 +1,7 @@
 ; 0-terminated string manipulation routines.
 
 string {
+    %option no_symbol_prefixing
 
     asmsub length(uword string @AY) clobbers(A) -> ubyte @Y {
         ; Returns the number of bytes in the string.
@@ -224,7 +225,7 @@ _done       rts
         }}
     }
 
-    asmsub lowerchar(ubyte char @A) -> ubyte @A {
+    asmsub lowerchar(ubyte character @A) -> ubyte @A {
         %asm {{
             and  #$7f
             cmp  #97
@@ -236,7 +237,7 @@ _done       rts
         }}
     }
 
-    asmsub upperchar(ubyte char @A) -> ubyte @A {
+    asmsub upperchar(ubyte character @A) -> ubyte @A {
         %asm {{
             cmp  #65
             bcc  +
@@ -279,11 +280,10 @@ _done       rts
 ;
 ; see http://6502.org/source/strings/patmatch.htm
 
-str = P8ZP_SCRATCH_W1
+strptr = P8ZP_SCRATCH_W1
 
-	stx  P8ZP_SCRATCH_REG
-	sta  str
-	sty  str+1
+	sta  strptr
+	sty  strptr+1
 	lda  cx16.r0
 	sta  modify_pattern1+1
 	sta  modify_pattern2+1
@@ -293,7 +293,6 @@ str = P8ZP_SCRATCH_W1
 	jsr  _match
 	lda  #0
 	rol  a
-	ldx  P8ZP_SCRATCH_REG
 	rts
 
 
@@ -307,9 +306,9 @@ next    lda $ffff,x   ; look at next pattern character    MODIFIED
 	iny             ; no, let's look at the string
 	cmp #'?'     ; is the pattern caracter a ques?
 	bne reg         ; no, it's a regular character
-	lda (str),y     ; yes, so it will match anything
+	lda (strptr),y     ; yes, so it will match anything
 	beq fail        ;  except the end of string
-reg     cmp (str),y     ; are both characters the same?
+reg     cmp (strptr),y     ; are both characters the same?
 	bne fail        ; no, so no match
 	inx             ; yes, keep checking
 	cmp #0          ; are we at end of string?
@@ -331,7 +330,7 @@ stloop  txa             ; we first try to match with * = ""
 	tax
 	bcs found       ; we found a match, return with c=1
 	iny             ; no match yet, try to grow * string
-	lda (str),y     ; are we at the end of string?
+	lda (strptr),y     ; are we at the end of string?
 	bne stloop      ; not yet, add a character
 fail    clc             ; yes, no match found, return with c=0
 	rts

compiler/res/prog8lib/test_stack.p8

@@ -1,12 +1,12 @@
-; utility debug code to print the X (evalstack) and SP (cpu stack) registers.
+; utility debug code to print the SP (cpu stack pointer) register.
 
 %import textio
 
 test_stack {
+    %option no_symbol_prefixing
 
     asmsub test() {
         %asm {{
-	stx  _saveX
 	lda  #13
 	jsr  txt.chrout
 	lda  #'-'
@@ -16,14 +16,6 @@ test_stack {
 	bne  -
 	lda  #13
 	jsr  txt.chrout
-	lda  #'x'
-	jsr  txt.chrout
-	lda  #'='
-	jsr  txt.chrout
-	lda  _saveX
-	jsr  txt.print_ub
-	lda  #' '
-	jsr  txt.chrout
 	lda  #'s'
 	jsr  txt.chrout
 	lda  #'p'
@@ -42,9 +34,7 @@ test_stack {
 	bne  -
 	lda  #13
 	jsr  txt.chrout
-	ldx  _saveX
 	rts
-_saveX	.byte 0
         }}
     }
 }

compiler/res/prog8lib/virtual/conv.p8

@@ -106,19 +106,21 @@ sub  str_uwhex  (uword value) {
 
 sub  str_uw0  (uword value) {
     ; ---- convert the uword in A/Y in decimal string form, with left padding 0s (5 positions total)
-    ubyte tenthousands = (value / 10000) as ubyte
-    value -= 10000*tenthousands
-    ubyte thousands = (value / 1000) as ubyte
-    value -= 1000*thousands
-    ubyte hundreds = (value / 100) as ubyte
-    value -= 100 as uword * hundreds
-    ubyte tens = (value / 10) as ubyte
-    value -= 10*tens
+    uword value2 = value/10
+    ubyte digits = value-value2*10 as ubyte
+    uword value3 = value2/10
+    ubyte tens = value2-value3*10 as ubyte
+    uword value4 = value3/10
+    ubyte hundreds = value3-value4*10 as ubyte
+    uword value5 = value4/10
+    ubyte thousands = value4-value5*10 as ubyte
+    uword value6 = value5/10
+    ubyte tenthousands = value5-value6*10 as ubyte
     string_out[0] = tenthousands+'0'
     string_out[1] = thousands+'0'
     string_out[2] = hundreds+'0'
     string_out[3] = tens+'0'
-    string_out[4] = value as ubyte + '0'
+    string_out[4] = digits+'0'
     string_out[5] = 0
 }
 
@@ -139,14 +141,16 @@ sub  str_w  (word value)  {
 }
 
 sub internal_str_uw(uword value, uword out_ptr) {
-    ubyte tenthousands = (value / 10000) as ubyte
-    value -= 10000*tenthousands
-    ubyte thousands = (value / 1000) as ubyte
-    value -= 1000*thousands
-    ubyte hundreds = (value / 100) as ubyte
-    value -= 100 as uword * hundreds
-    ubyte tens = (value / 10) as ubyte
-    value -= 10*tens
+    uword value2 = value/10
+    ubyte digits = value-value2*10 as ubyte
+    uword value3 = value2/10
+    ubyte tens = value2-value3*10 as ubyte
+    uword value4 = value3/10
+    ubyte hundreds = value3-value4*10 as ubyte
+    uword value5 = value4/10
+    ubyte thousands = value4-value5*10 as ubyte
+    uword value6 = value5/10
+    ubyte tenthousands = value5-value6*10 as ubyte
     if tenthousands
         goto output_tenthousands
     if thousands
@@ -169,7 +173,7 @@ output_tens:
     @(out_ptr) = tens+'0'
     out_ptr++
 output_ones:
-    @(out_ptr) = value as ubyte + '0'
+    @(out_ptr) = digits+'0'
     out_ptr++
     @(out_ptr) = 0
 }

compiler/res/prog8lib/virtual/diskio.p8

@@ -0,0 +1 @@
+; there is no diskio yet for the VM target.